\documentclass[10pt,a4paper]{article}

% Packages
\usepackage{fancyhdr}           % For header and footer
\usepackage{multicol}           % Allows multicols in tables
\usepackage{tabularx}           % Intelligent column widths
\usepackage{tabulary}           % Used in header and footer
\usepackage{hhline}             % Border under tables
\usepackage{graphicx}           % For images
\usepackage{xcolor}             % For hex colours
%\usepackage[utf8x]{inputenc}    % For unicode character support
\usepackage[T1]{fontenc}        % Without this we get weird character replacements
\usepackage{colortbl}           % For coloured tables
\usepackage{setspace}           % For line height
\usepackage{lastpage}           % Needed for total page number
\usepackage{seqsplit}           % Splits long words.
%\usepackage{opensans}          % Can't make this work so far. Shame. Would be lovely.
\usepackage[normalem]{ulem}     % For underlining links
% Most of the following are not required for the majority
% of cheat sheets but are needed for some symbol support.
\usepackage{amsmath}            % Symbols
\usepackage{MnSymbol}           % Symbols
\usepackage{wasysym}            % Symbols
%\usepackage[english,german,french,spanish,italian]{babel}              % Languages

% Document Info
\author{bhaskar}
\pdfinfo{
  /Title (data-science-101.pdf)
  /Creator (Cheatography)
  /Author (bhaskar)
  /Subject (Data Science 101 Cheat Sheet)
}

% Lengths and widths
\addtolength{\textwidth}{6cm}
\addtolength{\textheight}{-1cm}
\addtolength{\hoffset}{-3cm}
\addtolength{\voffset}{-2cm}
\setlength{\tabcolsep}{0.2cm} % Space between columns
\setlength{\headsep}{-12pt} % Reduce space between header and content
\setlength{\headheight}{85pt} % If less, LaTeX automatically increases it
\renewcommand{\footrulewidth}{0pt} % Remove footer line
\renewcommand{\headrulewidth}{0pt} % Remove header line
\renewcommand{\seqinsert}{\ifmmode\allowbreak\else\-\fi} % Hyphens in seqsplit
% This two commands together give roughly
% the right line height in the tables
\renewcommand{\arraystretch}{1.3}
\onehalfspacing

% Commands
\newcommand{\SetRowColor}[1]{\noalign{\gdef\RowColorName{#1}}\rowcolor{\RowColorName}} % Shortcut for row colour
\newcommand{\mymulticolumn}[3]{\multicolumn{#1}{>{\columncolor{\RowColorName}}#2}{#3}} % For coloured multi-cols
\newcolumntype{x}[1]{>{\raggedright}p{#1}} % New column types for ragged-right paragraph columns
\newcommand{\tn}{\tabularnewline} % Required as custom column type in use

% Font and Colours
\definecolor{HeadBackground}{HTML}{333333}
\definecolor{FootBackground}{HTML}{666666}
\definecolor{TextColor}{HTML}{333333}
\definecolor{DarkBackground}{HTML}{7086A3}
\definecolor{LightBackground}{HTML}{F6F7F9}
\renewcommand{\familydefault}{\sfdefault}
\color{TextColor}

% Header and Footer
\pagestyle{fancy}
\fancyhead{} % Set header to blank
\fancyfoot{} % Set footer to blank
\fancyhead[L]{
\noindent
\begin{multicols}{3}
\begin{tabulary}{5.8cm}{C}
    \SetRowColor{DarkBackground}
    \vspace{-7pt}
    {\parbox{\dimexpr\textwidth-2\fboxsep\relax}{\noindent
        \hspace*{-6pt}\includegraphics[width=5.8cm]{/web/www.cheatography.com/public/images/cheatography_logo.pdf}}
    }
\end{tabulary}
\columnbreak
\begin{tabulary}{11cm}{L}
    \vspace{-2pt}\large{\bf{\textcolor{DarkBackground}{\textrm{Data Science 101 Cheat Sheet}}}} \\
    \normalsize{by \textcolor{DarkBackground}{bhaskar} via \textcolor{DarkBackground}{\uline{cheatography.com/188368/cs/39288/}}}
\end{tabulary}
\end{multicols}}

\fancyfoot[L]{ \footnotesize
\noindent
\begin{multicols}{3}
\begin{tabulary}{5.8cm}{LL}
  \SetRowColor{FootBackground}
  \mymulticolumn{2}{p{5.377cm}}{\bf\textcolor{white}{Cheatographer}}  \\
  \vspace{-2pt}bhaskar \\
  \uline{cheatography.com/bhaskar} \\
  \end{tabulary}
\vfill
\columnbreak
\begin{tabulary}{5.8cm}{L}
  \SetRowColor{FootBackground}
  \mymulticolumn{1}{p{5.377cm}}{\bf\textcolor{white}{Cheat Sheet}}  \\
   \vspace{-2pt}Not Yet Published.\\
   Updated 20th June, 2023.\\
   Page {\thepage} of \pageref{LastPage}.
\end{tabulary}
\vfill
\columnbreak
\begin{tabulary}{5.8cm}{L}
  \SetRowColor{FootBackground}
  \mymulticolumn{1}{p{5.377cm}}{\bf\textcolor{white}{Sponsor}}  \\
  \SetRowColor{white}
  \vspace{-5pt}
  %\includegraphics[width=48px,height=48px]{dave.jpeg}
  Measure your website readability!\\
  www.readability-score.com
\end{tabulary}
\end{multicols}}




\begin{document}
\raggedright
\raggedcolumns

% Set font size to small. Switch to any value
% from this page to resize cheat sheet text:
% www.emerson.emory.edu/services/latex/latex_169.html
\footnotesize % Small font.

\begin{multicols*}{3}

\begin{tabularx}{5.377cm}{X}
\SetRowColor{DarkBackground}
\mymulticolumn{1}{x{5.377cm}}{\bf\textcolor{white}{Linear Regression Cheat Sheet}}  \tn
% Row 0
\SetRowColor{LightBackground}
\mymulticolumn{1}{x{5.377cm}}{Linear Regression Overview} \tn 
% Row Count 1 (+ 1)
% Row 1
\SetRowColor{white}
\mymulticolumn{1}{x{5.377cm}}{Linear regression is a statistical technique used to model the relationship between a dependent variable and one or more independent variables. \{\{nl\}\}\{\{nl\}\}It assumes a linear relationship between the independent variables and the dependent variable.} \tn 
% Row Count 6 (+ 5)
% Row 2
\SetRowColor{LightBackground}
\mymulticolumn{1}{x{5.377cm}}{Simple Linear Regression} \tn 
% Row Count 7 (+ 1)
% Row 3
\SetRowColor{white}
\mymulticolumn{1}{x{5.377cm}}{Simple linear regression involves a single independent variable (x) and a dependent variable (y) related by the equation: y = mx + c, where m is the slope and c is the intercept.} \tn 
% Row Count 11 (+ 4)
% Row 4
\SetRowColor{LightBackground}
\mymulticolumn{1}{x{5.377cm}}{Multiple Linear Regression} \tn 
% Row Count 12 (+ 1)
% Row 5
\SetRowColor{white}
\mymulticolumn{1}{x{5.377cm}}{Multiple linear regression involves more than one independent variable (x1, x2, x3, etc.) and a dependent variable (y) related by the equation: y = b0 + b1x1 + b2x2 + ... + bnxn, where b0 is the intercept, and b1, b2, ..., bn are the coefficients.} \tn 
% Row Count 17 (+ 5)
% Row 6
\SetRowColor{LightBackground}
\mymulticolumn{1}{x{5.377cm}}{Assumptions of Linear Regression} \tn 
% Row Count 18 (+ 1)
% Row 7
\SetRowColor{white}
\mymulticolumn{1}{x{5.377cm}}{Linearity: There should be a linear relationship between the independent and dependent variables. \{\{nl\}\}\{\{nl\}\}Independence: The observations should be independent of each other. \{\{nl\}\}\{\{nl\}\}Homoscedasticity: The variance of the residuals should be constant across all levels of the independent variables. \{\{nl\}\}\{\{nl\}\}Normality: The residuals should be normally distributed. \{\{nl\}\}\{\{nl\}\}No multicollinearity: The independent variables should not be highly correlated with each other.} \tn 
% Row Count 28 (+ 10)
% Row 8
\SetRowColor{LightBackground}
\mymulticolumn{1}{x{5.377cm}}{Fitting the Model} \tn 
% Row Count 29 (+ 1)
% Row 9
\SetRowColor{white}
\mymulticolumn{1}{x{5.377cm}}{The goal is to find the best-fitting line that minimizes the sum of squared residuals (differences between predicted and actual values). \{\{nl\}\}\{\{nl\}\}This is typically achieved using the method of least squares.} \tn 
% Row Count 34 (+ 5)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{X}
\SetRowColor{DarkBackground}
\mymulticolumn{1}{x{5.377cm}}{\bf\textcolor{white}{Linear Regression Cheat Sheet (cont)}}  \tn
% Row 10
\SetRowColor{LightBackground}
\mymulticolumn{1}{x{5.377cm}}{Interpreting Coefficients} \tn 
% Row Count 1 (+ 1)
% Row 11
\SetRowColor{white}
\mymulticolumn{1}{x{5.377cm}}{The intercept (b0) represents the expected value of the dependent variable when all independent variables are zero. \{\{nl\}\}\{\{nl\}\}The coefficients (b1, b2, ..., bn) represent the change in the dependent variable associated with a one-unit change in the corresponding independent variable, holding other variables constant.} \tn 
% Row Count 8 (+ 7)
% Row 12
\SetRowColor{LightBackground}
\mymulticolumn{1}{x{5.377cm}}{Evaluating Model Performance} \tn 
% Row Count 9 (+ 1)
% Row 13
\SetRowColor{white}
\mymulticolumn{1}{x{5.377cm}}{R-squared (R$^{\textrm{2}}$): Indicates the proportion of variance in the dependent variable explained by the independent variables. Higher values indicate a better fit. \{\{nl\}\}\{\{nl\}\}Adjusted R-squared: Similar to R-squared, but adjusts for the number of predictors in the model. \{\{nl\}\}\{\{nl\}\}Root Mean Squared Error (RMSE): Represents the average prediction error of the model. Lower values indicate better performance. \{\{nl\}\}\{\{nl\}\}Residual Analysis: Plotting residuals to check for patterns or outliers that violate assumptions.} \tn 
% Row Count 20 (+ 11)
% Row 14
\SetRowColor{LightBackground}
\mymulticolumn{1}{x{5.377cm}}{Handling Nonlinearity} \tn 
% Row Count 21 (+ 1)
% Row 15
\SetRowColor{white}
\mymulticolumn{1}{x{5.377cm}}{Polynomial Regression: Transforming the independent variables by adding polynomial terms (e.g., x\textasciicircum{}2, x\textasciicircum{}3) to capture nonlinear relationships. \{\{nl\}\}\{\{nl\}\}Logarithmic Transformation: Taking the logarithm of the dependent or independent variables to handle exponential growth or decay.} \tn 
% Row Count 27 (+ 6)
% Row 16
\SetRowColor{LightBackground}
\mymulticolumn{1}{x{5.377cm}}{Dealing with Multicollinearity} \tn 
% Row Count 28 (+ 1)
% Row 17
\SetRowColor{white}
\mymulticolumn{1}{x{5.377cm}}{Check Correlation: Identify highly correlated independent variables using correlation matrices or variance inflation factor (VIF) analysis. \{\{nl\}\}\{\{nl\}\}Remove or Combine Variables: Remove one of the highly correlated variables or combine them into a single variable.} \tn 
% Row Count 34 (+ 6)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{X}
\SetRowColor{DarkBackground}
\mymulticolumn{1}{x{5.377cm}}{\bf\textcolor{white}{Linear Regression Cheat Sheet (cont)}}  \tn
% Row 18
\SetRowColor{LightBackground}
\mymulticolumn{1}{x{5.377cm}}{Regularization Techniques} \tn 
% Row Count 1 (+ 1)
% Row 19
\SetRowColor{white}
\mymulticolumn{1}{x{5.377cm}}{Ridge Regression: Adds a penalty term to the sum of squared residuals to shrink the coefficients, reducing the impact of multicollinearity. \{\{nl\}\}\{\{nl\}\}Lasso Regression: Similar to Ridge regression, but with a penalty that can shrink coefficients to zero, effectively performing feature selection.} \tn 
% Row Count 7 (+ 6)
\hhline{>{\arrayrulecolor{DarkBackground}}-}
\end{tabularx}
\par\addvspace{1.3em}

\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Logistic Regression Cheat Sheet}}  \tn
% Row 0
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Logistic Regression Overview} \tn 
% Row Count 1 (+ 1)
% Row 1
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Logistic regression is a statistical technique used to model the relationship between a dependent variable and one or more independent variables. \{\{nl\}\}\{\{nl\}\}It is primarily used for binary classification problems, where the dependent variable takes on two categories} \tn 
% Row Count 7 (+ 6)
% Row 2
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Binary Logistic Regression} \tn 
% Row Count 8 (+ 1)
% Row 3
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Binary logistic regression involves a binary dependent variable (y) and one or more independent variables (x1, x2, x3, etc.). \{\{nl\}\}\{\{nl\}\}The logistic regression equation models the probability of the dependent variable belonging to a specific category.} \tn 
% Row Count 14 (+ 6)
% Row 4
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Logistic Regression Equation} \tn 
% Row Count 15 (+ 1)
% Row 5
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{The logistic regression equation is represented as: \{\{nl\}\}p = 1 / (1 + e\textasciicircum{}(-z)), where p is the probability of the event occurring, and z is the linear combination of the independent variables and their coefficients.} \tn 
% Row Count 20 (+ 5)
% Row 6
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Link Function} \tn 
% Row Count 21 (+ 1)
% Row 7
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Logistic regression uses the logistic or sigmoid function as the link function to map the linear combination of independent variables to a probability value between 0 and 1.} \tn 
% Row Count 25 (+ 4)
% Row 8
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Estimating Coefficients} \tn 
% Row Count 26 (+ 1)
% Row 9
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Coefficients are estimated using maximum likelihood estimation, which finds the values that maximize the likelihood of the observed data given the model. \{\{nl\}\}\{\{nl\}\}The coefficients represent the log-odds ratio, indicating the change in the log-odds of the event occurring for a one-unit change in the independent variable.} \tn 
% Row Count 33 (+ 7)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Logistic Regression Cheat Sheet (cont)}}  \tn
% Row 10
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Interpreting Coefficients} \tn 
% Row Count 1 (+ 1)
% Row 11
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{The coefficients can be exponentiated to obtain odds ratios, representing the change in odds of the event occurring for a one-unit change in the independent variable. \{\{nl\}\}\{\{nl\}\}Odds ratios greater than 1 indicate a positive association, while those less than 1 indicate a negative association.} \tn 
% Row Count 7 (+ 6)
% Row 12
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Evaluating Model Performance} \tn 
% Row Count 8 (+ 1)
% Row 13
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Accuracy: The proportion of correctly classified instances. \{\{nl\}\}\{\{nl\}\}Confusion Matrix: A table showing the true positives, true negatives, false positives, and false negatives. \{\{nl\}\}\{\{nl\}\}Precision: The proportion of true positives out of all positive predictions (TP / (TP + FP)). \{\{nl\}\}\{\{nl\}\}Recall (Sensitivity): The proportion of true positives out of all actual positives (TP / (TP + FN)). \{\{nl\}\}\{\{nl\}\}Specificity: The proportion of true negatives out of all actual negatives (TN / (TN + FP)). \{\{nl\}\}\{\{nl\}\}F1 Score: A measure that combines precision and recall to balance their importance.} \tn 
% Row Count 20 (+ 12)
% Row 14
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Regularization Techniques} \tn 
% Row Count 21 (+ 1)
% Row 15
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Ridge Regression (L2 regularization): Adds a penalty term to the loss function to shrink the coefficients, reducing overfitting. \{\{nl\}\}\{\{nl\}\}Lasso Regression (L1 regularization): Similar to Ridge regression but can shrink coefficients to zero, effectively performing feature selection.} \tn 
% Row Count 27 (+ 6)
% Row 16
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Multiclass Logistic Regression} \tn 
% Row Count 28 (+ 1)
% Row 17
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Multiclass logistic regression extends binary logistic regression to handle more than two categories. \{\{nl\}\}\{\{nl\}\}One-vs-Rest (OvR) or One-vs-All (OvA) is a common approach where separate binary logistic regression models are trained for each class against the rest.} \tn 
% Row Count 34 (+ 6)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Logistic Regression Cheat Sheet (cont)}}  \tn
% Row 18
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Dealing with Imbalanced Data} \tn 
% Row Count 1 (+ 1)
% Row 19
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Adjust Class Weights: Assign higher weights to the minority class to address the class imbalance during model training. \{\{nl\}\}\{\{nl\}\}Resampling Techniques: Oversampling the minority class or undersampling the majority class to create a balanced dataset.} \tn 
% Row Count 7 (+ 6)
\hhline{>{\arrayrulecolor{DarkBackground}}--}
\end{tabularx}
\par\addvspace{1.3em}

\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{k-Nearest Neighbors Cheat Sheet}}  \tn
% Row 0
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{k-Nearest Neighbors Overview} \tn 
% Row Count 1 (+ 1)
% Row 1
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{k-Nearest Neighbors is a non-parametric and instance-based machine learning algorithm used for classification and regression tasks. \{\{nl\}\}\{\{nl\}\}It predicts the class or value of a new data point based on the majority vote or average of its k nearest neighbors in the feature space.} \tn 
% Row Count 7 (+ 6)
% Row 2
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Choosing k} \tn 
% Row Count 8 (+ 1)
% Row 3
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{The value of k represents the number of nearest neighbors to consider when making predictions. \{\{nl\}\}\{\{nl\}\}A small value of k (e.g., 1) may lead to overfitting, while a large value of k may lead to oversimplification and loss of local patterns. \{\{nl\}\}\{\{nl\}\}The optimal value of k is typically determined through hyperparameter tuning using techniques like cross-validation} \tn 
% Row Count 16 (+ 8)
% Row 4
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Distance Metrics} \tn 
% Row Count 17 (+ 1)
% Row 5
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Euclidean Distance: Calculates the straight-line distance between two points in the feature space. \{\{nl\}\}\{\{nl\}\}Manhattan Distance: Calculates the sum of absolute differences between the coordinates of two points. \{\{nl\}\}\{\{nl\}\}Other distance metrics like Minkowski, Cosine, and Hamming distance can also be used depending on the data type and problem domain.} \tn 
% Row Count 25 (+ 8)
% Row 6
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Feature Scaling} \tn 
% Row Count 26 (+ 1)
% Row 7
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{It's crucial to scale the features before applying k-NN, as it is sensitive to the scale of the features. \{\{nl\}\}\{\{nl\}\}Standardization (mean = 0, standard deviation = 1) or normalization (scaling to a range) techniques like min-max scaling are commonly used.} \tn 
% Row Count 32 (+ 6)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{k-Nearest Neighbors Cheat Sheet (cont)}}  \tn
% Row 8
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Handling Categorical Features} \tn 
% Row Count 1 (+ 1)
% Row 9
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Categorical features must be encoded into numerical values before applying k-NN. \{\{nl\}\}\{\{nl\}\}One-Hot Encoding: Creates binary dummy variables for each category, representing their presence or absence. \{\{nl\}\}\{\{nl\}\}Label Encoding: Assigns a unique numerical label to each category.} \tn 
% Row Count 7 (+ 6)
% Row 10
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Classifying New Instances} \tn 
% Row Count 8 (+ 1)
% Row 11
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{For classification tasks, the class of a new instance is determined by the majority class among its k nearest neighbors. \{\{nl\}\}\{\{nl\}\}Voting Mechanisms: Simple majority vote, weighted vote (based on distance or confidence), or distance-weighted vote (inverse of distance) can be used.} \tn 
% Row Count 14 (+ 6)
% Row 12
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Regression with k-NN} \tn 
% Row Count 15 (+ 1)
% Row 13
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{For regression tasks, the predicted value of a new instance is typically the average (mean or median) of the target values of its k nearest neighbors.} \tn 
% Row Count 18 (+ 3)
% Row 14
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Model Evaluation} \tn 
% Row Count 19 (+ 1)
% Row 15
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Accuracy: Proportion of correctly classified instances for classification tasks. \{\{nl\}\}\{\{nl\}\}Mean Squared Error (MSE): Average of the squared differences between the predicted and actual values for regression tasks. \{\{nl\}\}\{\{nl\}\}Cross-Validation: Technique to assess the performance of the k-NN model by splitting the data into multiple folds.} \tn 
% Row Count 26 (+ 7)
% Row 16
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Curse of Dimensionality} \tn 
% Row Count 27 (+ 1)
% Row 17
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{As the number of features increases, the feature space becomes increasingly sparse, making k-NN less effective. \{\{nl\}\}\{\{nl\}\}Feature selection or dimensionality reduction techniques (e.g., Principal Component Analysis) can help mitigate this issue.} \tn 
% Row Count 32 (+ 5)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{k-Nearest Neighbors Cheat Sheet (cont)}}  \tn
% Row 18
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Advantages and Limitations} \tn 
% Row Count 1 (+ 1)
% Row 19
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Advantages: Simplicity, no assumptions about data distribution, and ability to capture complex patterns. \{\{nl\}\}\{\{nl\}\}Limitations: Computationally expensive for large datasets, sensitivity to feature scaling, and inability to handle missing values well.} \tn 
% Row Count 7 (+ 6)
\hhline{>{\arrayrulecolor{DarkBackground}}--}
\end{tabularx}
\par\addvspace{1.3em}

\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Support Vector Machines Cheet Sheat}}  \tn
% Row 0
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Support Vector Machines Overview} \tn 
% Row Count 1 (+ 1)
% Row 1
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Support Vector Machines is a supervised machine learning algorithm used for classification and regression tasks. \{\{nl\}\}\{\{nl\}\}It finds an optimal hyperplane that maximally separates or fits the data points in the feature space.} \tn 
% Row Count 6 (+ 5)
% Row 2
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Linear SVM} \tn 
% Row Count 7 (+ 1)
% Row 3
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Linear SVM constructs a linear decision boundary to separate data points of different classes. \{\{nl\}\}\{\{nl\}\}It aims to maximize the margin, which is the perpendicular distance between the decision boundary and the nearest data points (support vectors).} \tn 
% Row Count 13 (+ 6)
% Row 4
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Kernel Trick} \tn 
% Row Count 14 (+ 1)
% Row 5
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{The kernel trick allows SVMs to efficiently handle non-linearly separable data by mapping the data to a higher-dimensional feature space. \{\{nl\}\}\{\{nl\}\}Common kernel functions include Linear, Polynomial, Radial Basis Function (RBF), and Sigmoid.} \tn 
% Row Count 19 (+ 5)
% Row 6
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Soft Margin SVM} \tn 
% Row Count 20 (+ 1)
% Row 7
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Soft Margin SVM allows for some misclassification in order to achieve a more flexible decision boundary. \{\{nl\}\}\{\{nl\}\}It introduces a regularization parameter (C) to control the trade-off between maximizing the margin and minimizing misclassification.} \tn 
% Row Count 25 (+ 5)
% Row 8
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Choosing the Right Kernel} \tn 
% Row Count 26 (+ 1)
% Row 9
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Linear Kernel: Suitable for linearly separable data or when the number of features is large compared to the number of samples. \{\{nl\}\}\{\{nl\}\}Polynomial Kernel: Suitable for problems with intermediate complexity and higher-order polynomial relationships. \{\{nl\}\}\{\{nl\}\}RBF Kernel: Suitable for complex and non-linear relationships; the most commonly used kernel. \{\{nl\}\}\{\{nl\}\}Sigmoid Kernel: Suitable for problems influenced by logistic regression or neural networks.} \tn 
% Row Count 36 (+ 10)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Support Vector Machines Cheet Sheat (cont)}}  \tn
% Row 10
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Model Training and Optimization} \tn 
% Row Count 1 (+ 1)
% Row 11
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{SVM training involves solving a quadratic programming problem to find the optimal hyperplane. \{\{nl\}\}\{\{nl\}\}The optimization process can be computationally expensive for large datasets, but various optimization techniques (e.g., Sequential Minimal Optimization) can improve efficiency.} \tn 
% Row Count 7 (+ 6)
% Row 12
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Tuning Parameters} \tn 
% Row Count 8 (+ 1)
% Row 13
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{C (Regularization Parameter): Controls the trade-off between misclassification and the width of the margin. A smaller C allows more misclassification, while a larger C enforces stricter classification. \{\{nl\}\}\{\{nl\}\}Gamma (Kernel Coefficient): Influences the shape of the decision boundary. A higher gamma value leads to a more complex decision boundary.} \tn 
% Row Count 16 (+ 8)
% Row 14
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Multi-Class Classification} \tn 
% Row Count 17 (+ 1)
% Row 15
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{One-vs-Rest (OvR) or One-vs-One (OvO) strategies can be used to extend SVM to multi-class classification problems. \{\{nl\}\}\{\{nl\}\}OvR: Trains separate binary classifiers for each class against the rest. \{\{nl\}\}\{\{nl\}\}OvO: Trains a binary classifier for every pair of classes} \tn 
% Row Count 23 (+ 6)
% Row 16
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Handling Imbalanced Data} \tn 
% Row Count 24 (+ 1)
% Row 17
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Class imbalance can affect SVM performance. Techniques such as resampling (undersampling or oversampling) and adjusting class weights can help address this issue.} \tn 
% Row Count 28 (+ 4)
% Row 18
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Advantages and Limitations} \tn 
% Row Count 29 (+ 1)
% Row 19
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Advantages: Effective in high-dimensional spaces, robust against overfitting, and suitable for both linear and non-linear classification. \{\{nl\}\}\{\{nl\}\}Limitations: Computationally intensive for large datasets, sensitive to hyperparameter tuning, and challenging to interpret complex models.} \tn 
% Row Count 35 (+ 6)
\hhline{>{\arrayrulecolor{DarkBackground}}--}
\end{tabularx}
\par\addvspace{1.3em}

\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Decision Tree Cheat Sheet}}  \tn
% Row 0
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Decision Tree Overview} \tn 
% Row Count 1 (+ 1)
% Row 1
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Decision Trees are a supervised machine learning algorithm used for classification and regression tasks. \{\{nl\}\}\{\{nl\}\}They learn a hierarchical structure of decisions and conditions from the data to make predictions.} \tn 
% Row Count 6 (+ 5)
% Row 2
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Tree Construction} \tn 
% Row Count 7 (+ 1)
% Row 3
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Decision Trees are constructed through a top-down, recursive partitioning process called recursive binary splitting. \{\{nl\}\}\{\{nl\}\}The algorithm selects the best feature at each node to split the data based on certain criteria (e.g., information gain, Gini impurity).} \tn 
% Row Count 13 (+ 6)
% Row 4
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Splitting Criteria} \tn 
% Row Count 14 (+ 1)
% Row 5
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Information Gain: Measures the reduction in entropy (or increase in information) achieved by splitting on a particular feature. \{\{nl\}\}\{\{nl\}\}Gini Impurity: Measures the probability of misclassifying a randomly chosen element if it were labeled randomly according to the class distribution.} \tn 
% Row Count 20 (+ 6)
% Row 6
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Handling Continuous and Categorical Features} \tn 
% Row Count 21 (+ 1)
% Row 7
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{For continuous features, decision tree algorithms use threshold values to split the data. \{\{nl\}\}\{\{nl\}\}For categorical features, each category forms a separate branch in the decision tree.} \tn 
% Row Count 25 (+ 4)
% Row 8
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Tree Pruning} \tn 
% Row Count 26 (+ 1)
% Row 9
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Pruning is a technique used to avoid overfitting by reducing the complexity of the decision tree. \{\{nl\}\}\{\{nl\}\}Pre-pruning: Setting constraints on tree depth, minimum samples per leaf, or maximum number of leaf nodes during tree construction. \{\{nl\}\}\{\{nl\}\}Post-pruning: Removing or collapsing branches that provide little information gain or result in minimal improvements in performance.} \tn 
% Row Count 34 (+ 8)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Decision Tree Cheat Sheet (cont)}}  \tn
% Row 10
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Handling Missing Values} \tn 
% Row Count 1 (+ 1)
% Row 11
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Decision Trees can handle missing values by treating them as a separate category or by imputing missing values before tree construction.} \tn 
% Row Count 4 (+ 3)
% Row 12
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Handling Imbalanced Data} \tn 
% Row Count 5 (+ 1)
% Row 13
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Imbalanced class distributions can bias the decision tree. Techniques like class weighting, undersampling, or oversampling can help address this issue.} \tn 
% Row Count 9 (+ 4)
% Row 14
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Feature Importance} \tn 
% Row Count 10 (+ 1)
% Row 15
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Decision Trees provide feature importance scores based on how much each feature contributes to the overall split decisions. \{\{nl\}\}\{\{nl\}\}Importance can be measured by the total reduction in impurity or the total information gain associated with a feature} \tn 
% Row Count 16 (+ 6)
% Row 16
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Ensemble Methods} \tn 
% Row Count 17 (+ 1)
% Row 17
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Random Forest: An ensemble of decision trees where each tree is trained on a random subset of the data with replacement. It reduces overfitting and improves performance. \{\{nl\}\}\{\{nl\}\}Gradient Boosting: Builds an ensemble by sequentially adding decision trees, with each tree correcting the mistakes made by the previous trees.} \tn 
% Row Count 24 (+ 7)
% Row 18
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Advantages and Limitations} \tn 
% Row Count 25 (+ 1)
% Row 19
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Advantages: Easy to understand and interpret, handles both numerical and categorical data, and can capture non-linear relationships. \{\{nl\}\}\{\{nl\}\}Limitations: Prone to overfitting, sensitive to small changes in data, and may not generalize well to unseen data if the tree structure is too complex.} \tn 
% Row Count 31 (+ 6)
\hhline{>{\arrayrulecolor{DarkBackground}}--}
\end{tabularx}
\par\addvspace{1.3em}

\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Random Forest Cheat Sheet}}  \tn
% Row 0
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Random Forest Overview} \tn 
% Row Count 1 (+ 1)
% Row 1
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Random Forest is an ensemble learning algorithm that combines multiple decision trees to make predictions. \{\{nl\}\}\{\{nl\}\}It is used for both classification and regression tasks and improves upon the individual decision trees' performance and robustness.} \tn 
% Row Count 7 (+ 6)
% Row 2
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Ensemble of Decision Trees} \tn 
% Row Count 8 (+ 1)
% Row 3
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Random Forest creates an ensemble by constructing a set of decision trees on random subsets of the training data (bootstrap sampling). \{\{nl\}\}\{\{nl\}\}Each decision tree is trained independently, making predictions based on majority voting (classification) or averaging (regression) of the individual tree predictions.} \tn 
% Row Count 15 (+ 7)
% Row 4
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Random Feature Subsets} \tn 
% Row Count 16 (+ 1)
% Row 5
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{In addition to using random subsets of the training data, Random Forest also considers a random subset of features at each node for constructing the decision trees. \{\{nl\}\}\{\{nl\}\}This randomness reduces the correlation between trees and promotes diversity, leading to improved generalization.} \tn 
% Row Count 22 (+ 6)
% Row 6
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Building Decision Trees} \tn 
% Row Count 23 (+ 1)
% Row 7
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Each decision tree in the Random Forest is constructed using a subset of the training data and a subset of the available features. \{\{nl\}\}\{\{nl\}\}Tree construction follows the usual process of recursive binary splitting based on criteria like information gain or Gini impurity.} \tn 
% Row Count 29 (+ 6)
% Row 8
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Feature Importance} \tn 
% Row Count 30 (+ 1)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Random Forest Cheat Sheet (cont)}}  \tn
% Row 9
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Random Forest provides a measure of feature importance based on how much each feature contributes to the ensemble's predictive performance. \{\{nl\}\}\{\{nl\}\}Importance can be calculated by evaluating the average decrease in impurity or the average decrease in a split criterion (e.g., Gini index) caused by a feature.} \tn 
% Row Count 7 (+ 7)
% Row 10
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Out-of-Bag (OOB) Error} \tn 
% Row Count 8 (+ 1)
% Row 11
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Random Forest uses the out-of-bag samples (not included in the bootstrap sample) to estimate the model's performance without the need for cross-validation. \{\{nl\}\}\{\{nl\}\}OOB error provides a good estimate of the model's generalization performance and can be used for model evaluation and hyperparameter tuning.} \tn 
% Row Count 15 (+ 7)
% Row 12
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Hyperparameter Tuning} \tn 
% Row Count 16 (+ 1)
% Row 13
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Important hyperparameters to consider when working with Random Forests include the number of trees (n\_estimators), maximum depth of each tree (max\_depth), minimum samples required to split a node (min\_samples\_split), and maximum number of features to consider for each split (max\_features).} \tn 
% Row Count 22 (+ 6)
% Row 14
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Handling Imbalanced Data} \tn 
% Row Count 23 (+ 1)
% Row 15
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Random Forests can handle imbalanced data by adjusting class weights during tree construction or by using sampling techniques like oversampling the minority class or undersampling the majority class.} \tn 
% Row Count 27 (+ 4)
% Row 16
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Advantages and Limitations} \tn 
% Row Count 28 (+ 1)
% Row 17
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Advantages: Robust to overfitting, can handle high-dimensional data, provides feature importance, and performs well on various types of problems. \{\{nl\}\}\{\{nl\}\}Limitations: Requires more computational resources than individual decision trees, can be slower to train and predict, and may not perform well on extremely imbalanced datasets.} \tn 
% Row Count 35 (+ 7)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Random Forest Cheat Sheet (cont)}}  \tn
% Row 18
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Applications} \tn 
% Row Count 1 (+ 1)
% Row 19
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Random Forests are commonly used in various domains, including classification tasks such as image recognition, text classification, fraud detection, and regression tasks like predicting housing prices or stock market trends.} \tn 
% Row Count 6 (+ 5)
\hhline{>{\arrayrulecolor{DarkBackground}}--}
\end{tabularx}
\par\addvspace{1.3em}

\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Gradient Boosting Cheat Sheet}}  \tn
% Row 0
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Gradient Boosting Overview} \tn 
% Row Count 1 (+ 1)
% Row 1
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Gradient Boosting is an ensemble learning algorithm that combines multiple weak prediction models (typically decision trees) to create a strong predictive model. \{\{nl\}\}\{\{nl\}\}It is used for both classification and regression tasks and sequentially improves the model's performance by minimizing the errors of the previous models.} \tn 
% Row Count 8 (+ 7)
% Row 2
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Boosting Process} \tn 
% Row Count 9 (+ 1)
% Row 3
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Gradient Boosting builds the ensemble by adding decision trees sequentially, with each subsequent tree correcting the mistakes of the previous ones. \{\{nl\}\}\{\{nl\}\}The trees are built in a greedy manner, minimizing a loss function (e.g., mean squared error for regression, log loss for classification) at each step.} \tn 
% Row Count 16 (+ 7)
% Row 4
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Gradient Descent} \tn 
% Row Count 17 (+ 1)
% Row 5
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Gradient Boosting optimizes the loss function using gradient descent. \{\{nl\}\}\{\{nl\}\}The model calculates the negative gradient of the loss function with respect to the current model's predictions and fits a new weak learner to this gradient.} \tn 
% Row Count 22 (+ 5)
% Row 6
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Learning Rate and Number of Trees} \tn 
% Row Count 23 (+ 1)
% Row 7
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{The learning rate (shrinkage factor) controls the contribution of each tree to the ensemble. A smaller learning rate requires more trees for convergence but can lead to better generalization. \{\{nl\}\}\{\{nl\}\}The number of trees (iterations) determines the complexity of the model and affects both training time and the risk of overfitting.} \tn 
% Row Count 30 (+ 7)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Gradient Boosting Cheat Sheet (cont)}}  \tn
% Row 8
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Regularization Techniques} \tn 
% Row Count 1 (+ 1)
% Row 9
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Regularization is applied to control the complexity of the model and avoid overfitting. \{\{nl\}\}\{\{nl\}\}Tree Depth: Restricting the maximum depth of each tree can prevent overfitting and speed up training. \{\{nl\}\}\{\{nl\}\}Tree Pruning: Applying pruning techniques to remove branches with little contribution to the model's performance.} \tn 
% Row Count 8 (+ 7)
% Row 10
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Feature Subsampling} \tn 
% Row Count 9 (+ 1)
% Row 11
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Gradient Boosting can use random feature subsets similar to Random Forests to introduce randomness and increase diversity among the weak learners. \{\{nl\}\}\{\{nl\}\}It can prevent overfitting when dealing with high-dimensional data or datasets with a large number of features.} \tn 
% Row Count 15 (+ 6)
% Row 12
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Handling Imbalanced Data} \tn 
% Row Count 16 (+ 1)
% Row 13
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Techniques such as class weighting or sampling (undersampling the majority class or oversampling the minority class) can be applied to address imbalanced datasets during Gradient Boosting.} \tn 
% Row Count 20 (+ 4)
% Row 14
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Hyperparameter Tuning} \tn 
% Row Count 21 (+ 1)
% Row 15
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Important hyperparameters to consider when working with Gradient Boosting include the learning rate, number of trees, maximum depth of each tree, and regularization parameters like subsample and colsample\_bytree.} \tn 
% Row Count 26 (+ 5)
% Row 16
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Early Stopping} \tn 
% Row Count 27 (+ 1)
% Row 17
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Early stopping is a technique used to prevent overfitting and speed up training by monitoring the model's performance on a validation set. \{\{nl\}\}\{\{nl\}\}Training stops when the performance on the validation set does not improve for a specified number of iterations.} \tn 
% Row Count 33 (+ 6)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Gradient Boosting Cheat Sheet (cont)}}  \tn
% Row 18
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Applications} \tn 
% Row Count 1 (+ 1)
% Row 19
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Gradient Boosting has been successfully applied to a wide range of tasks, including web search ranking, anomaly detection, click-through rate prediction, and personalized medicine.} \tn 
% Row Count 5 (+ 4)
\hhline{>{\arrayrulecolor{DarkBackground}}--}
\end{tabularx}
\par\addvspace{1.3em}

\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Naive Bayes Cheat Sheet}}  \tn
% Row 0
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Naive Bayes Overview} \tn 
% Row Count 1 (+ 1)
% Row 1
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Naive Bayes is a probabilistic machine learning algorithm based on Bayes' theorem with the assumption of independence between features. \{\{nl\}\}\{\{nl\}\}It is primarily used for classification tasks and is efficient, simple, and often works well in practice} \tn 
% Row Count 7 (+ 6)
% Row 2
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Bayes' Theorem} \tn 
% Row Count 8 (+ 1)
% Row 3
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Bayes' theorem calculates the posterior probability of a class given the observed evidence. \{\{nl\}\}\{\{nl\}\}P(Class|Features) = (P(Features|Class) * P(Class)) / P(Features)} \tn 
% Row Count 12 (+ 4)
% Row 4
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Assumption of Feature Independence} \tn 
% Row Count 13 (+ 1)
% Row 5
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Naive Bayes assumes that the features are conditionally independent given the class label, which is a simplifying assumption to make the calculations more tractable. \{\{nl\}\}\{\{nl\}\}Despite this assumption rarely being true in reality, Naive Bayes can still perform well in practice} \tn 
% Row Count 19 (+ 6)
% Row 6
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Types of Naive Bayes Classifiers} \tn 
% Row Count 20 (+ 1)
% Row 7
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Gaussian Naive Bayes: Assumes a Gaussian distribution for continuous features and estimates the mean and variance for each class. \{\{nl\}\}\{\{nl\}\}Multinomial Naive Bayes: Suitable for discrete features, typically used for text classification tasks, where features represent word frequencies. \{\{nl\}\}\{\{nl\}\}Bernoulli Naive Bayes: Similar to multinomial, but assumes binary features (presence or absence).} \tn 
% Row Count 28 (+ 8)
% Row 8
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Feature Probability Estimation} \tn 
% Row Count 29 (+ 1)
% Row 9
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{For continuous features, Gaussian Naive Bayes estimates the mean and variance for each class. \{\{nl\}\}\{\{nl\}\}For discrete features, Multinomial Naive Bayes estimates the probability of each feature occurring in each class. \{\{nl\}\}\{\{nl\}\}For binary features, Bernoulli Naive Bayes estimates the probability of each feature being present in each class.} \tn 
% Row Count 36 (+ 7)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Naive Bayes Cheat Sheet (cont)}}  \tn
% Row 10
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Handling Zero Probabilities} \tn 
% Row Count 1 (+ 1)
% Row 11
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{The Naive Bayes classifier may encounter zero probabilities if a particular feature does not occur in the training set for a specific class. \{\{nl\}\}\{\{nl\}\}To handle this, techniques like Laplace smoothing or add-one smoothing can be applied to avoid zero probabilities.} \tn 
% Row Count 7 (+ 6)
% Row 12
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Handling Continuous Features} \tn 
% Row Count 8 (+ 1)
% Row 13
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Gaussian Naive Bayes assumes a Gaussian distribution for continuous features. \{\{nl\}\}\{\{nl\}\}Continuous features can be discretized into bins or transformed into categorical variables before using Naive Bayes.} \tn 
% Row Count 13 (+ 5)
% Row 14
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Text Classification with Naive Bayes} \tn 
% Row Count 14 (+ 1)
% Row 15
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Naive Bayes is commonly used for text classification tasks, such as spam detection or sentiment analysis. \{\{nl\}\}\{\{nl\}\}Text data is typically preprocessed by tokenization, removing stop words, and applying techniques like TF-IDF or Bag-of-Words representation before using Naive Bayes.} \tn 
% Row Count 20 (+ 6)
% Row 16
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Advantages and Limitations} \tn 
% Row Count 21 (+ 1)
% Row 17
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Advantages: Simplicity, efficiency, and can handle high-dimensional data well. \{\{nl\}\}\{\{nl\}\}Limitations: Strong independence assumption may not hold in reality, and it can be sensitive to irrelevant features. It may struggle with rare or unseen combinations of features.} \tn 
% Row Count 27 (+ 6)
% Row 18
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Handling Imbalanced Data} \tn 
% Row Count 28 (+ 1)
% Row 19
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Naive Bayes can face challenges with imbalanced datasets where the class distribution is skewed. \{\{nl\}\}\{\{nl\}\}Techniques like class weighting or resampling (undersampling or oversampling) can help alleviate the impact of imbalanced data.} \tn 
% Row Count 33 (+ 5)
\hhline{>{\arrayrulecolor{DarkBackground}}--}
\end{tabularx}
\par\addvspace{1.3em}

\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Principal Component Analysis Cheat Sheet}}  \tn
% Row 0
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{PCA Overview} \tn 
% Row Count 1 (+ 1)
% Row 1
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{PCA is a dimensionality reduction technique used to transform a high-dimensional dataset into a lower-dimensional space. \{\{nl\}\}\{\{nl\}\}It identifies the principal components, which are orthogonal directions that capture the maximum variance in the data.} \tn 
% Row Count 7 (+ 6)
% Row 2
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Variance and Covariance} \tn 
% Row Count 8 (+ 1)
% Row 3
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{PCA is based on the variance-covariance matrix or the correlation matrix of the dataset. \{\{nl\}\}\{\{nl\}\}Variance measures the spread of data along a specific axis, while covariance measures the relationship between two variables.} \tn 
% Row Count 13 (+ 5)
% Row 4
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Steps in PCA} \tn 
% Row Count 14 (+ 1)
% Row 5
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Standardize the data: PCA works best with standardized data to ensure equal importance across different variables. \{\{nl\}\}\{\{nl\}\}Calculate the covariance matrix or correlation matrix: This represents the relationships between the variables in the dataset. \{\{nl\}\}\{\{nl\}\}Compute the eigenvectors and eigenvalues: These eigenvectors represent the principal components, and the corresponding eigenvalues indicate the amount of variance explained by each component. \{\{nl\}\}\{\{nl\}\}Select the desired number of principal components: Choose the top components that explain the majority of the variance in the data. \{\{nl\}\}\{\{nl\}\}Transform the data: Project the original data onto the selected principal components to obtain the lower-dimensional representation.} \tn 
% Row Count 29 (+ 15)
% Row 6
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Explained Variance and Scree Plot} \tn 
% Row Count 30 (+ 1)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Principal Component Analysis Cheat Sheet (cont)}}  \tn
% Row 7
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Explained variance ratio indicates the proportion of variance explained by each principal component. \{\{nl\}\}\{\{nl\}\}A scree plot visualizes the explained variance ratio for each component, helping to determine the number of components to retain.} \tn 
% Row Count 5 (+ 5)
% Row 8
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Dimensionality Reduction and Reconstruction} \tn 
% Row Count 6 (+ 1)
% Row 9
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{PCA reduces the dimensionality of the dataset by selecting a subset of principal components. \{\{nl\}\}\{\{nl\}\}Reconstruction of the original data is possible by projecting the lower-dimensional representation back into the original feature space.} \tn 
% Row Count 11 (+ 5)
% Row 10
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Applications of PCA} \tn 
% Row Count 12 (+ 1)
% Row 11
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Dimensionality reduction: PCA can help visualize high-dimensional data, reduce noise, and eliminate redundant or correlated features. \{\{nl\}\}\{\{nl\}\}Data compression: PCA can compress the data by retaining only the most important components. \{\{nl\}\}\{\{nl\}\}Feature extraction: PCA can extract meaningful features from complex data, facilitating subsequent analysis.} \tn 
% Row Count 20 (+ 8)
% Row 12
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Interpretation of Principal Components} \tn 
% Row Count 21 (+ 1)
% Row 13
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Principal components are linear combinations of the original features. \{\{nl\}\}\{\{nl\}\}The direction of a principal component represents the most significant variation in the data. \{\{nl\}\}\{\{nl\}\}The magnitude of the component's loading on a particular feature indicates its contribution to that component.} \tn 
% Row Count 27 (+ 6)
% Row 14
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Assumptions and Limitations} \tn 
% Row Count 28 (+ 1)
% Row 15
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{PCA assumes linear relationships between variables and requires variables to be continuous or approximately continuous. \{\{nl\}\}\{\{nl\}\}It may not be suitable for datasets with nonlinear relationships or when interpretability of individual features is essential.} \tn 
% Row Count 34 (+ 6)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Principal Component Analysis Cheat Sheet (cont)}}  \tn
% Row 16
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Extensions to PCA} \tn 
% Row Count 1 (+ 1)
% Row 17
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Kernel PCA: An extension that allows nonlinear transformations of the data. \{\{nl\}\}\{\{nl\}\}Sparse PCA: A variant that encourages sparsity in the loadings, resulting in a more interpretable representation.} \tn 
% Row Count 6 (+ 5)
% Row 18
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Implementation and Libraries} \tn 
% Row Count 7 (+ 1)
% Row 19
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{PCA is implemented in various programming languages. Commonly used libraries include scikit-learn (Python), caret (R), and numpy (Python) for numerical computations.} \tn 
% Row Count 11 (+ 4)
\hhline{>{\arrayrulecolor{DarkBackground}}--}
\end{tabularx}
\par\addvspace{1.3em}

\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Cluster Analysis Cheat Sheet}}  \tn
% Row 0
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Cluster Analysis Overview} \tn 
% Row Count 1 (+ 1)
% Row 1
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Cluster Analysis is an unsupervised learning technique used to group similar objects or data points into clusters based on their characteristics or proximity. \{\{nl\}\}\{\{nl\}\}It helps discover hidden patterns, similarities, or structures within the data.} \tn 
% Row Count 6 (+ 5)
% Row 2
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Types of Cluster Analysis} \tn 
% Row Count 7 (+ 1)
% Row 3
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Hierarchical Clustering: Builds a hierarchy of clusters by recursively merging or splitting clusters based on a similarity measure. \{\{nl\}\}\{\{nl\}\}K-means Clustering: Divides the data into a predetermined number (k) of non-overlapping clusters by minimizing the within-cluster sum of squares. \{\{nl\}\}\{\{nl\}\}Density-based Clustering: Groups data points based on density and identifies regions with higher density as clusters. \{\{nl\}\}\{\{nl\}\}Model-based Clustering: Assumes a specific statistical model for each cluster and estimates model parameters to assign data points to clusters.} \tn 
% Row Count 19 (+ 12)
% Row 4
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Similarity and Distance Measures} \tn 
% Row Count 20 (+ 1)
% Row 5
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Cluster analysis often relies on similarity or distance measures to determine the proximity between data points. \{\{nl\}\}\{\{nl\}\}Common distance measures include Euclidean distance, Manhattan distance, and cosine similarity.} \tn 
% Row Count 25 (+ 5)
% Row 6
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Hierarchical Clustering} \tn 
% Row Count 26 (+ 1)
% Row 7
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Agglomerative (Bottom-Up): Starts with each data point as a separate cluster and iteratively merges the closest pairs of clusters until all points belong to a single cluster. \{\{nl\}\}\{\{nl\}\}Divisive (Top-Down): Begins with all data points in one cluster and recursively splits clusters until each data point is in its own cluster.} \tn 
% Row Count 33 (+ 7)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Cluster Analysis Cheat Sheet (cont)}}  \tn
% Row 8
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{K-means Clustering} \tn 
% Row Count 1 (+ 1)
% Row 9
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Randomly initializes k cluster centroids, assigns each data point to the nearest centroid, recalculates the centroids based on the mean of assigned points, and repeats until convergence. \{\{nl\}\}\{\{nl\}\}The choice of the number of clusters (k) is important and can impact the results.} \tn 
% Row Count 7 (+ 6)
% Row 10
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Density-based Clustering (DBSCAN)} \tn 
% Row Count 8 (+ 1)
% Row 11
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Density-based Spatial Clustering of Applications with Noise (DBSCAN) groups data points based on density and identifies core points, border points, and noise points. \{\{nl\}\}\{\{nl\}\}It defines clusters as dense regions separated by sparser areas and does not require specifying the number of clusters in advance.} \tn 
% Row Count 15 (+ 7)
% Row 12
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Model-based Clustering (Gaussian Mixture Models)} \tn 
% Row Count 16 (+ 1)
% Row 13
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Gaussian Mixture Models (GMM) assume that the data points are generated from a mixture of Gaussian distributions. \{\{nl\}\}\{\{nl\}\}It estimates the parameters of the Gaussian distributions and assigns data points to clusters based on the likelihood.} \tn 
% Row Count 21 (+ 5)
% Row 14
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Evaluation of Clustering} \tn 
% Row Count 22 (+ 1)
% Row 15
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Internal Evaluation: Measures the quality of clustering using intrinsic criteria such as the silhouette coefficient or within-cluster sum of squares. \{\{nl\}\}\{\{nl\}\}External Evaluation: Compares the clustering results to a known ground truth, if available, using external criteria like purity or F-measure.} \tn 
% Row Count 29 (+ 7)
% Row 16
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Handling Missing Data and Outliers} \tn 
% Row Count 30 (+ 1)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Cluster Analysis Cheat Sheet (cont)}}  \tn
% Row 17
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Missing data can be handled by imputation techniques before clustering. \{\{nl\}\}\{\{nl\}\}Outliers can significantly impact clustering results. Techniques like outlier detection or preprocessing methods can be applied to mitigate their influence.} \tn 
% Row Count 5 (+ 5)
% Row 18
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Visualization of Clustering Results} \tn 
% Row Count 6 (+ 1)
% Row 19
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Dimensionality reduction techniques like PCA or t-SNE can be used to visualize high-dimensional clustering results in lower-dimensional space. \{\{nl\}\}\{\{nl\}\}Scatter plots, heatmaps, or dendrograms can provide insights into the clustering structure.} \tn 
% Row Count 11 (+ 5)
\hhline{>{\arrayrulecolor{DarkBackground}}--}
\end{tabularx}
\par\addvspace{1.3em}

\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Neural Networks Cheat Sheet}}  \tn
% Row 0
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Neural Network Basics} \tn 
% Row Count 1 (+ 1)
% Row 1
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Neural networks are a class of machine learning models inspired by the human brain's structure and functioning. \{\{nl\}\}\{\{nl\}\}They consist of interconnected nodes called neurons, organized in layers (input, hidden, and output).} \tn 
% Row Count 6 (+ 5)
% Row 2
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Activation Functions} \tn 
% Row Count 7 (+ 1)
% Row 3
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Activation functions introduce non-linearity to the neural network and help model complex relationships. \{\{nl\}\}\{\{nl\}\}Common activation functions include sigmoid, tanh, ReLU, and softmax (for multiclass classification).} \tn 
% Row Count 12 (+ 5)
% Row 4
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Forward Propagation} \tn 
% Row Count 13 (+ 1)
% Row 5
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Forward propagation is the process of passing input data through the neural network to obtain predictions. \{\{nl\}\}\{\{nl\}\}Each neuron applies a weighted sum of inputs, followed by the activation function, to produce an output.} \tn 
% Row Count 18 (+ 5)
% Row 6
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Loss Functions} \tn 
% Row Count 19 (+ 1)
% Row 7
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Loss functions quantify the difference between predicted outputs and true labels. \{\{nl\}\}\{\{nl\}\}Regression: Mean Squared Error (MSE), Mean Absolute Error (MAE). \{\{nl\}\}\{\{nl\}\}Binary Classification: Binary Cross-Entropy. \{\{nl\}\}\{\{nl\}\}Multiclass Classification: Categorical Cross-Entropy.} \tn 
% Row Count 25 (+ 6)
% Row 8
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Backpropagation} \tn 
% Row Count 26 (+ 1)
% Row 9
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Backpropagation is used to update the weights of the neural network based on the calculated gradients of the loss function. \{\{nl\}\}\{\{nl\}\}It propagates the error from the output layer to the previous layers, adjusting the weights through gradient descent.} \tn 
% Row Count 32 (+ 6)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Neural Networks Cheat Sheet (cont)}}  \tn
% Row 10
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Gradient Descent Optimization} \tn 
% Row Count 1 (+ 1)
% Row 11
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Gradient Descent is an optimization algorithm used to minimize the loss function and update the weights iteratively. \{\{nl\}\}\{\{nl\}\}Common variants include Stochastic Gradient Descent (SGD), Mini-Batch Gradient Descent, and Adam.} \tn 
% Row Count 6 (+ 5)
% Row 12
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Regularization Techniques} \tn 
% Row Count 7 (+ 1)
% Row 13
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Regularization helps prevent overfitting and improves the generalization of the neural network. \{\{nl\}\}\{\{nl\}\}Common techniques include L1 and L2 regularization (weight decay), dropout, and early stopping.} \tn 
% Row Count 12 (+ 5)
% Row 14
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Hyperparameter Tuning} \tn 
% Row Count 13 (+ 1)
% Row 15
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Neural networks have various hyperparameters that need to be tuned for optimal performance. \{\{nl\}\}\{\{nl\}\}Examples include learning rate, number of layers, number of neurons per layer, batch size, and activation functions.} \tn 
% Row Count 18 (+ 5)
% Row 16
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Convolutional Neural Networks (CNN)} \tn 
% Row Count 19 (+ 1)
% Row 17
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{CNNs are specialized neural networks commonly used for image and video processing tasks. \{\{nl\}\}\{\{nl\}\}They consist of convolutional layers, pooling layers, and fully connected layers, exploiting the spatial structure of data.} \tn 
% Row Count 24 (+ 5)
% Row 18
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Recurrent Neural Networks (RNN)} \tn 
% Row Count 25 (+ 1)
% Row 19
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{RNNs are designed for sequential data processing tasks, such as natural language processing and time series analysis. \{\{nl\}\}\{\{nl\}\}They have recurrent connections that allow information to persist and flow across different time steps.} \tn 
% Row Count 30 (+ 5)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Neural Networks Cheat Sheet (cont)}}  \tn
% Row 20
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Transfer Learning} \tn 
% Row Count 1 (+ 1)
% Row 21
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Transfer learning leverages pre-trained neural network models on large datasets for similar tasks to improve performance on smaller datasets. \{\{nl\}\}\{\{nl\}\}By using pre-trained models as a starting point, training time can be reduced, and generalization can be enhanced.} \tn 
% Row Count 7 (+ 6)
% Row 22
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Hardware Acceleration} \tn 
% Row Count 8 (+ 1)
% Row 23
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{To speed up training and inference, specialized hardware like GPUs (Graphics Processing Units) or TPUs (Tensor Processing Units) can be utilized.} \tn 
% Row Count 11 (+ 3)
\hhline{>{\arrayrulecolor{DarkBackground}}--}
\end{tabularx}
\par\addvspace{1.3em}

\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Convolutional Neural Networks Cheat Sheet}}  \tn
% Row 0
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Convolutional Neural Networks Overview} \tn 
% Row Count 1 (+ 1)
% Row 1
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{CNNs are a type of neural network specifically designed for processing grid-like data, such as images. \{\{nl\}\}\{\{nl\}\}They leverage the concept of convolution to extract relevant features from the input data.} \tn 
% Row Count 6 (+ 5)
% Row 2
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Convolutional Layers} \tn 
% Row Count 7 (+ 1)
% Row 3
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Convolutional layers perform the main feature extraction in CNNs. \{\{nl\}\}\{\{nl\}\}Each layer consists of multiple filters (also called kernels) that scan the input data through convolution operations. \{\{nl\}\}\{\{nl\}\}Convolution applies a sliding window over the input and performs element-wise multiplication and summing to produce feature maps.} \tn 
% Row Count 14 (+ 7)
% Row 4
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Pooling Layers} \tn 
% Row Count 15 (+ 1)
% Row 5
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Pooling layers reduce the spatial dimensions of the feature maps, reducing computational complexity and providing spatial invariance. \{\{nl\}\}\{\{nl\}\}Common types of pooling include Max Pooling (selecting the maximum value in each pooling region) and Average Pooling (taking the average).} \tn 
% Row Count 21 (+ 6)
% Row 6
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Activation Functions} \tn 
% Row Count 22 (+ 1)
% Row 7
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Activation functions introduce non-linearity to the CNN and enable modeling complex relationships. \{\{nl\}\}\{\{nl\}\}ReLU (Rectified Linear Unit) is commonly used as the activation function in CNNs, promoting faster convergence and avoiding the vanishing gradient problem.} \tn 
% Row Count 28 (+ 6)
% Row 8
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Fully Connected Layers} \tn 
% Row Count 29 (+ 1)
% Row 9
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Fully connected layers, also known as dense layers, are traditional neural network layers where each neuron is connected to every neuron in the previous layer. \{\{nl\}\}\{\{nl\}\}They provide the final classification or regression output by combining the learned features.} \tn 
% Row Count 35 (+ 6)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Convolutional Neural Networks Cheat Sheet (cont)}}  \tn
% Row 10
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Loss Functions} \tn 
% Row Count 1 (+ 1)
% Row 11
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Loss functions quantify the difference between predicted outputs and true labels in CNNs. \{\{nl\}\}\{\{nl\}\}Common loss functions include Mean Squared Error (MSE) for regression tasks and Cross-Entropy for classification tasks.} \tn 
% Row Count 6 (+ 5)
% Row 12
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Training Techniques} \tn 
% Row Count 7 (+ 1)
% Row 13
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{CNNs are typically trained using backpropagation and gradient descent optimization methods. \{\{nl\}\}\{\{nl\}\}Techniques like Dropout (randomly deactivating neurons during training) and Batch Normalization (normalizing inputs to accelerate training) are commonly used to improve generalization and performance.} \tn 
% Row Count 14 (+ 7)
% Row 14
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Data Augmentation} \tn 
% Row Count 15 (+ 1)
% Row 15
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Data augmentation techniques help increase the diversity of the training data by applying transformations such as rotations, translations, flips, or scaling. \{\{nl\}\}\{\{nl\}\}This helps improve the model's ability to generalize and reduces overfitting.} \tn 
% Row Count 20 (+ 5)
% Row 16
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Transfer Learning} \tn 
% Row Count 21 (+ 1)
% Row 17
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Transfer learning leverages pretrained CNN models on large datasets and adapts them to new tasks or smaller datasets. \{\{nl\}\}\{\{nl\}\}Pretrained models like VGGNet and ResNet are available, allowing transfer of learned features to new applications.} \tn 
% Row Count 26 (+ 5)
% Row 18
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Object Localization and Detection} \tn 
% Row Count 27 (+ 1)
% Row 19
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{CNNs can be extended to perform object localization and detection tasks using techniques like bounding box regression and region proposal networks (RPN).} \tn 
% Row Count 31 (+ 4)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Convolutional Neural Networks Cheat Sheet (cont)}}  \tn
% Row 20
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Semantic Segmentation} \tn 
% Row Count 1 (+ 1)
% Row 21
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Semantic segmentation assigns a label to each pixel or region in an image, allowing detailed object-level understanding. \{\{nl\}\}\{\{nl\}\}Fully Convolutional Networks (FCNs) are commonly used for semantic segmentation.} \tn 
% Row Count 6 (+ 5)
% Row 22
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Hardware Acceleration} \tn 
% Row Count 7 (+ 1)
% Row 23
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{CNNs can benefit from specialized hardware like GPUs (Graphics Processing Units) or TPUs (Tensor Processing Units) for faster training and inference.} \tn 
% Row Count 10 (+ 3)
\hhline{>{\arrayrulecolor{DarkBackground}}--}
\end{tabularx}
\par\addvspace{1.3em}

\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Recurrent Neural Networks Cheat Sheet}}  \tn
% Row 0
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Recurrent Neural Network (RNN) Basics} \tn 
% Row Count 1 (+ 1)
% Row 1
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{RNNs are a class of neural networks designed for processing sequential data, such as time series, natural language, and speech. \{\{nl\}\}\{\{nl\}\}They have recurrent connections that allow information to persist and flow across different time steps} \tn 
% Row Count 6 (+ 5)
% Row 2
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{RNN Cell} \tn 
% Row Count 7 (+ 1)
% Row 3
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{The basic building block of an RNN is the RNN cell, which maintains a hidden state and takes input at each time step. \{\{nl\}\}\{\{nl\}\}The hidden state captures the memory of past inputs and influences future predictions.} \tn 
% Row Count 12 (+ 5)
% Row 4
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Vanishing and Exploding Gradients} \tn 
% Row Count 13 (+ 1)
% Row 5
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{RNNs can suffer from the vanishing gradient problem, where gradients diminish exponentially as they propagate through time, leading to difficulties in learning long-term dependencies. \{\{nl\}\}\{\{nl\}\}Conversely, exploding gradients can occur when gradients grow rapidly during backpropagation.} \tn 
% Row Count 19 (+ 6)
% Row 6
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Long Short-Term Memory (LSTM)} \tn 
% Row Count 20 (+ 1)
% Row 7
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{LSTMs are a type of RNN that address the vanishing gradient problem by using gating mechanisms. \{\{nl\}\}\{\{nl\}\}They introduce memory cells, input gates, output gates, and forget gates to selectively remember or forget information.} \tn 
% Row Count 25 (+ 5)
% Row 8
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Gated Recurrent Unit (GRU)} \tn 
% Row Count 26 (+ 1)
% Row 9
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{GRUs are another type of RNN that address the vanishing gradient problem and have a simpler architecture compared to LSTMs. \{\{nl\}\}\{\{nl\}\}They use reset and update gates to control the flow of information through the network.} \tn 
% Row Count 31 (+ 5)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Recurrent Neural Networks Cheat Sheet (cont)}}  \tn
% Row 10
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Bidirectional RNNs} \tn 
% Row Count 1 (+ 1)
% Row 11
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Bidirectional RNNs process the input sequence in both forward and backward directions, capturing information from past and future contexts. \{\{nl\}\}\{\{nl\}\}They are useful when the current prediction depends on both past and future context.} \tn 
% Row Count 6 (+ 5)
% Row 12
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Sequence-to-Sequence Models} \tn 
% Row Count 7 (+ 1)
% Row 13
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Sequence-to-sequence models, often built with RNNs, are used for tasks such as machine translation, text summarization, and speech recognition. \{\{nl\}\}\{\{nl\}\}They encode the input sequence into a fixed-size representation (context vector) and decode it to generate the output sequence.} \tn 
% Row Count 13 (+ 6)
% Row 14
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Attention Mechanism} \tn 
% Row Count 14 (+ 1)
% Row 15
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Attention mechanisms enhance the capability of RNNs by selectively focusing on different parts of the input sequence. \{\{nl\}\}\{\{nl\}\}They assign different weights to each input element, emphasizing more relevant information during decoding or generating.} \tn 
% Row Count 20 (+ 6)
% Row 16
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Training and Backpropagation Through Time (BPTT)} \tn 
% Row Count 21 (+ 1)
% Row 17
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{RNNs are trained using BPTT, which extends backpropagation to handle sequences. \{\{nl\}\}\{\{nl\}\}BPTT unfolds the RNN through time, allowing error gradients to be calculated and applied to update the weights.} \tn 
% Row Count 26 (+ 5)
% Row 18
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Applications of RNNs} \tn 
% Row Count 27 (+ 1)
% Row 19
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Language modeling, text generation, and sentiment analysis. \{\{nl\}\}\{\{nl\}\}Machine translation and natural language understanding. \{\{nl\}\}\{\{nl\}\}Speech recognition and speech synthesis. \{\{nl\}\}Time series forecasting and anomaly detection.} \tn 
% Row Count 32 (+ 5)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Recurrent Neural Networks Cheat Sheet (cont)}}  \tn
% Row 20
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Handling Variable-Length Inputs} \tn 
% Row Count 1 (+ 1)
% Row 21
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Techniques like padding, masking, and sequence bucketing can be used to handle inputs of different lengths in RNNs.} \tn 
% Row Count 4 (+ 3)
% Row 22
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Hardware Acceleration} \tn 
% Row Count 5 (+ 1)
% Row 23
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{RNNs, especially LSTMs and GRUs, can benefit from specialized hardware like GPUs (Graphics Processing Units) or TPUs (Tensor Processing Units) for faster training and inference.} \tn 
% Row Count 9 (+ 4)
\hhline{>{\arrayrulecolor{DarkBackground}}--}
\end{tabularx}
\par\addvspace{1.3em}

\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Generative Adversarial Networks Cheat Sheet}}  \tn
% Row 0
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Generative Adversarial Networks (GAN) Basics} \tn 
% Row Count 1 (+ 1)
% Row 1
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{GANs are a class of deep learning models composed of two components: a generator and a discriminator. \{\{nl\}\}\{\{nl\}\}The generator learns to generate synthetic data samples that resemble real data, while the discriminator tries to distinguish between real and fake samples.} \tn 
% Row Count 7 (+ 6)
% Row 2
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Generator} \tn 
% Row Count 8 (+ 1)
% Row 3
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{The generator takes random noise as input and generates synthetic samples. \{\{nl\}\}\{\{nl\}\}It typically consists of one or more layers of neural networks, often using transpose convolutions for upsampling.} \tn 
% Row Count 13 (+ 5)
% Row 4
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Discriminator} \tn 
% Row Count 14 (+ 1)
% Row 5
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{The discriminator takes a sample as input and estimates the probability of it being real or fake. \{\{nl\}\}\{\{nl\}\}It typically consists of one or more layers of neural networks, often using convolutions for feature extraction.} \tn 
% Row Count 19 (+ 5)
% Row 6
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Adversarial Training} \tn 
% Row Count 20 (+ 1)
% Row 7
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{The generator and discriminator are trained in an adversarial manner. \{\{nl\}\}\{\{nl\}\}The generator tries to generate samples that fool the discriminator, while the discriminator aims to correctly classify real and fake samples.} \tn 
% Row Count 25 (+ 5)
% Row 8
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Loss Functions} \tn 
% Row Count 26 (+ 1)
% Row 9
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{The generator and discriminator are trained using different loss functions. \{\{nl\}\}\{\{nl\}\}The generator's loss function encourages the generated samples to be classified as real by the discriminator. \{\{nl\}\}The discriminator's loss function penalizes misclassifying real and fake samples.} \tn 
% Row Count 32 (+ 6)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Generative Adversarial Networks Cheat Sheet (cont)}}  \tn
% Row 10
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Mode Collapse} \tn 
% Row Count 1 (+ 1)
% Row 11
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Mode collapse occurs when the generator produces limited and repetitive samples, failing to capture the diversity of the real data distribution. \{\{nl\}\}\{\{nl\}\}Techniques like minibatch discrimination and feature matching can help alleviate mode collapse.} \tn 
% Row Count 7 (+ 6)
% Row 12
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Deep Convolutional GAN (DCGAN)} \tn 
% Row Count 8 (+ 1)
% Row 13
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{DCGAN is a popular GAN architecture that uses convolutional neural networks for both the generator and discriminator. \{\{nl\}\}\{\{nl\}\}It leverages convolutional and transpose convolutional layers to generate and discriminate images.} \tn 
% Row Count 13 (+ 5)
% Row 14
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Conditional GAN (cGAN)} \tn 
% Row Count 14 (+ 1)
% Row 15
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{cGANs introduce additional information (such as class labels) to guide the generation process. \{\{nl\}\}\{\{nl\}\}The generator and discriminator take both random noise and conditional information as input.} \tn 
% Row Count 18 (+ 4)
% Row 16
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Evaluation of GANs} \tn 
% Row Count 19 (+ 1)
% Row 17
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Evaluating GANs is challenging as there is no direct objective function to optimize. \{\{nl\}\}\{\{nl\}\}Common evaluation methods include visual inspection, Inception Score, Fréchet Inception Distance (FID), and Precision and Recall curves.} \tn 
% Row Count 24 (+ 5)
% Row 18
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Unsupervised Representation Learning} \tn 
% Row Count 25 (+ 1)
% Row 19
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{GANs can learn meaningful representations without explicit labels. \{\{nl\}\}\{\{nl\}\}By training on a large unlabeled dataset, the generator can capture and generate high-level features.} \tn 
% Row Count 29 (+ 4)
% Row 20
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Variational Autoencoder (VAE) vs. GAN} \tn 
% Row Count 30 (+ 1)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Generative Adversarial Networks Cheat Sheet (cont)}}  \tn
% Row 21
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{VAEs and GANs are both generative models but differ in their underlying principles. \{\{nl\}\}\{\{nl\}\}VAEs focus on learning latent representations and reconstruction, while GANs emphasize generating realistic samples.} \tn 
% Row Count 5 (+ 5)
% Row 22
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Applications of GANs} \tn 
% Row Count 6 (+ 1)
% Row 23
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Image synthesis and generation. \{\{nl\}\}\{\{nl\}\}Style transfer and image-to-image translation. \{\{nl\}\}\{\{nl\}\}Data augmentation and synthesis for training other models. \{\{nl\}\}\{\{nl\}\}Text-to-image synthesis and generation.} \tn 
% Row Count 11 (+ 5)
\hhline{>{\arrayrulecolor{DarkBackground}}--}
\end{tabularx}
\par\addvspace{1.3em}

\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Transfer Learning Cheat Sheet}}  \tn
% Row 0
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{What is Transfer Learning?} \tn 
% Row Count 1 (+ 1)
% Row 1
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Transfer learning is a technique in machine learning where knowledge gained from training one model is applied to another related task or dataset. \{\{nl\}\}\{\{nl\}\}It leverages pre-trained models and their learned representations to improve performance and reduce the need for extensive training on new datasets.} \tn 
% Row Count 8 (+ 7)
% Row 2
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Benefits of Transfer Learning} \tn 
% Row Count 9 (+ 1)
% Row 3
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Reduces the need for large labeled datasets for training new models. \{\{nl\}\}\{\{nl\}\}Saves computational resources and time required for training. \{\{nl\}\}\{\{nl\}\}Helps generalize learned features to new tasks or domains. \{\{nl\}\}\{\{nl\}\}Improves model performance, especially with limited data.} \tn 
% Row Count 15 (+ 6)
% Row 4
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Popular Pre-trained Models} \tn 
% Row Count 16 (+ 1)
% Row 5
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Image Classification: VGG, ResNet, Inception, MobileNet, EfficientNet. \{\{nl\}\}\{\{nl\}\}Natural Language Processing: Word2Vec, GloVe, BERT, GPT, Transformer.} \tn 
% Row Count 20 (+ 4)
% Row 6
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Steps for Transfer Learning} \tn 
% Row Count 21 (+ 1)
% Row 7
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Select a pre-trained model: Choose a model that was trained on a large dataset and is suitable for your task. \{\{nl\}\}\{\{nl\}\}Remove the top layers: Remove the final layers responsible for task-specific predictions. \{\{nl\}\}\{\{nl\}\}Feature Extraction: Extract features from the pre-trained model by passing your dataset through the remaining layers. \{\{nl\}\}\{\{nl\}\}Add new layers: Add new layers to the pre-trained model to adapt it to your specific task. \{\{nl\}\}\{\{nl\}\}Train the new model: Fine-tune the new layers with your labeled dataset while keeping the pre-trained weights fixed or updating them with a smaller learning rate. \{\{nl\}\}\{\{nl\}\}Evaluate and Iterate: Evaluate the performance of your model on a validation set and iterate on the architecture or hyperparameters if necessary.} \tn 
% Row Count 37 (+ 16)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Transfer Learning Cheat Sheet (cont)}}  \tn
% Row 8
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Transfer Learning Techniques} \tn 
% Row Count 1 (+ 1)
% Row 9
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Feature Extraction: Extract high-level features from the pre-trained model and add new layers for task-specific predictions. \{\{nl\}\}\{\{nl\}\}Fine-tuning: Fine-tune the pre-trained model's weights by updating them during training with a smaller learning rate.} \tn 
% Row Count 7 (+ 6)
% Row 10
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Data Augmentation} \tn 
% Row Count 8 (+ 1)
% Row 11
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Apply data augmentation techniques such as rotation, translation, scaling, flipping, or cropping to increase the diversity of your training data. \{\{nl\}\}\{\{nl\}\}Data augmentation helps prevent overfitting and improves generalization.} \tn 
% Row Count 13 (+ 5)
% Row 12
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Domain Adaptation} \tn 
% Row Count 14 (+ 1)
% Row 13
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Domain adaptation is a form of transfer learning where the source and target domains differ, requiring adjustments to make the model generalize well. \{\{nl\}\}\{\{nl\}\}Techniques like adversarial training, self-training, or domain-specific fine-tuning can be used for domain adaptation.} \tn 
% Row Count 20 (+ 6)
% Row 14
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Choosing Layers for Transfer} \tn 
% Row Count 21 (+ 1)
% Row 15
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Earlier layers in a pre-trained model learn low-level features like edges and textures, while later layers learn high-level features. \{\{nl\}\}\{\{nl\}\}For small datasets, it's often beneficial to use earlier layers for transfer, as they capture more general features.} \tn 
% Row Count 27 (+ 6)
% Row 16
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Size of Training Data} \tn 
% Row Count 28 (+ 1)
% Row 17
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{The size of the new dataset influences the amount of transfer learning required. \{\{nl\}\}\{\{nl\}\}With limited data, it's crucial to rely more on the pre-trained weights and perform minimal fine-tuning to avoid overfitting.} \tn 
% Row Count 33 (+ 5)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Transfer Learning Cheat Sheet (cont)}}  \tn
% Row 18
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Transfer Learning in Different Domains} \tn 
% Row Count 1 (+ 1)
% Row 19
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Transfer learning is applicable across various domains, including computer vision, natural language processing, audio processing, and more. \{\{nl\}\}\{\{nl\}\}The choice of pre-trained models and the techniques used may vary based on the specific domain.} \tn 
% Row Count 6 (+ 5)
% Row 20
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Avoiding Negative Transfer} \tn 
% Row Count 7 (+ 1)
% Row 21
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Negative transfer occurs when the knowledge from the source task hinders the performance on the target task. \{\{nl\}\}\{\{nl\}\}It can be mitigated by selecting a source task that is related or has shared underlying patterns with the target task.} \tn 
% Row Count 12 (+ 5)
% Row 22
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Model Evaluation} \tn 
% Row Count 13 (+ 1)
% Row 23
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Evaluate the performance of the transfer learning model using appropriate metrics for your specific task, such as accuracy, precision, recall, F1-score, or mean squared error.} \tn 
% Row Count 17 (+ 4)
\hhline{>{\arrayrulecolor{DarkBackground}}--}
\end{tabularx}
\par\addvspace{1.3em}

\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Reinforcement Learning Cheat Sheet}}  \tn
% Row 0
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Reinforcement Learning Basics} \tn 
% Row Count 1 (+ 1)
% Row 1
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{RL is a branch of machine learning where an agent learns to interact with an environment to maximize a reward signal. \{\{nl\}\}\{\{nl\}\}The agent learns through a trial-and-error process, taking actions and receiving feedback from the environment.} \tn 
% Row Count 6 (+ 5)
% Row 2
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Key Components} \tn 
% Row Count 7 (+ 1)
% Row 3
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Agent: The learner or decision-maker that interacts with the environment. \{\{nl\}\}\{\{nl\}\}Environment: The external system with which the agent interacts. \{\{nl\}\}\{\{nl\}\}State: The current representation of the environment at a particular time step. \{\{nl\}\}\{\{nl\}\}Action: The decision or choice made by the agent based on the state. \{\{nl\}\}\{\{nl\}\}Reward: The feedback signal that the agent receives from the environment after taking an action.} \tn 
% Row Count 16 (+ 9)
% Row 4
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Markov Decision Process (MDP)} \tn 
% Row Count 17 (+ 1)
% Row 5
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{MDP provides a mathematical framework for modeling RL problems with states, actions, rewards, and state transitions. \{\{nl\}\}\{\{nl\}\}It assumes the Markov property, where the future state depends only on the current state and action, disregarding the history} \tn 
% Row Count 23 (+ 6)
% Row 6
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Value Function} \tn 
% Row Count 24 (+ 1)
% Row 7
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{The value function estimates the expected return or cumulative reward an agent will receive from a particular state or state-action pair. \{\{nl\}\}\{\{nl\}\}Value functions can be represented as state-value functions (V(s)) or action-value functions (Q(s, a)).} \tn 
% Row Count 30 (+ 6)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Reinforcement Learning Cheat Sheet (cont)}}  \tn
% Row 8
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Policy} \tn 
% Row Count 1 (+ 1)
% Row 9
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{The policy determines the agent's behavior, mapping states to actions. \{\{nl\}\}\{\{nl\}\}It can be deterministic or stochastic, providing the agent's action selection strategy.} \tn 
% Row Count 5 (+ 4)
% Row 10
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Exploration vs. Exploitation} \tn 
% Row Count 6 (+ 1)
% Row 11
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Exploration refers to the agent's search for new actions or states to gather more information about the environment. \{\{nl\}\}\{\{nl\}\}Exploitation refers to the agent's tendency to choose actions that are expected to yield the highest immediate rewards based on its current knowledge.} \tn 
% Row Count 12 (+ 6)
% Row 12
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Temporal Difference (TD) Learning} \tn 
% Row Count 13 (+ 1)
% Row 13
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{TD learning is a method for updating value functions based on the difference between the estimated and actual rewards received. \{\{nl\}\}\{\{nl\}\}Q-learning and SARSA are popular TD learning algorithms.} \tn 
% Row Count 17 (+ 4)
% Row 14
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Policy Gradient Methods} \tn 
% Row Count 18 (+ 1)
% Row 15
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Policy gradient methods directly optimize the policy by updating its parameters based on the gradients of expected rewards. \{\{nl\}\}\{\{nl\}\}They use techniques like REINFORCE, Proximal Policy Optimization (PPO), and Trust Region Policy Optimization (TRPO).} \tn 
% Row Count 24 (+ 6)
% Row 16
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Exploration Techniques} \tn 
% Row Count 25 (+ 1)
% Row 17
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Epsilon-Greedy: Randomly selects a random action with a probability epsilon to encourage exploration. \{\{nl\}\}\{\{nl\}\}Upper Confidence Bound (UCB): Balances exploration and exploitation using an optimistic value estimate. \{\{nl\}\}\{\{nl\}\}Thompson Sampling: Selects actions based on random samples from the posterior distribution of action values.} \tn 
% Row Count 32 (+ 7)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Reinforcement Learning Cheat Sheet (cont)}}  \tn
% Row 18
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Deep Reinforcement Learning (DRL)} \tn 
% Row Count 1 (+ 1)
% Row 19
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{DRL combines RL with deep neural networks to handle high-dimensional state spaces and complex tasks. \{\{nl\}\}\{\{nl\}\}Deep Q-Networks (DQN) and Deep Deterministic Policy Gradient (DDPG) are popular DRL algorithms.} \tn 
% Row Count 6 (+ 5)
% Row 20
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Off-Policy vs. On-Policy} \tn 
% Row Count 7 (+ 1)
% Row 21
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Off-policy methods learn the value function or policy using data collected from a different policy. \{\{nl\}\}\{\{nl\}\}On-policy methods learn from the direct interaction of the agent with the environment.} \tn 
% Row Count 11 (+ 4)
% Row 22
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Model-Based vs. Model-Free} \tn 
% Row Count 12 (+ 1)
% Row 23
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Model-based methods learn a model of the environment to plan and make decisions. \{\{nl\}\}\{\{nl\}\}Model-free methods directly learn the optimal policy or value function without explicitly modeling the environment dynamics.} \tn 
% Row Count 17 (+ 5)
\hhline{>{\arrayrulecolor{DarkBackground}}--}
\end{tabularx}
\par\addvspace{1.3em}

\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Time Series Forecasting Cheat Sheet}}  \tn
% Row 0
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Time Series Basics} \tn 
% Row Count 1 (+ 1)
% Row 1
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Time series data is a sequence of observations collected over time, typically at regular intervals. \{\{nl\}\}\{\{nl\}\}It exhibits temporal dependencies, trends, seasonality, and may contain noise.} \tn 
% Row Count 5 (+ 4)
% Row 2
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Stationarity} \tn 
% Row Count 6 (+ 1)
% Row 3
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Stationary time series have constant mean, variance, and autocovariance over time. \{\{nl\}\}\{\{nl\}\}Stationarity is desirable for accurate forecasting.} \tn 
% Row Count 9 (+ 3)
% Row 4
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Trends and Seasonality} \tn 
% Row Count 10 (+ 1)
% Row 5
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Trend refers to the long-term upward or downward movement in a time series. \{\{nl\}\}\{\{nl\}\}Seasonality refers to patterns that repeat at fixed intervals. \{\{nl\}\}\{\{nl\}\}Identifying and handling trends and seasonality is important for accurate forecasting.} \tn 
% Row Count 15 (+ 5)
% Row 6
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Autocorrelation Function (ACF) and Partial Autocorrelation Function (PACF)} \tn 
% Row Count 17 (+ 2)
% Row 7
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{ACF measures the correlation between a time series and its lagged values. \{\{nl\}\}\{\{nl\}\}PACF measures the correlation between a time series and its lagged values, excluding the intermediate lags. \{\{nl\}\}They help identify the order of autoregressive (AR) and moving average (MA) components in time series models.} \tn 
% Row Count 24 (+ 7)
% Row 8
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Time Series Models} \tn 
% Row Count 25 (+ 1)
% Row 9
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Autoregressive Integrated Moving Average (ARIMA): A linear model that combines AR and MA components to handle stationary time series. \{\{nl\}\}\{\{nl\}\}Seasonal ARIMA (SARIMA): Extends ARIMA to handle seasonal time series data. \{\{nl\}\}Exponential Smoothing Methods: Models that assign exponentially decreasing weights to past observations. \{\{nl\}\}\{\{nl\}\}Prophet: An additive regression model that captures trend, seasonality, and holiday effects. \{\{nl\}\}\{\{nl\}\}Vector Autoregression (VAR): A multivariate time series model that captures the relationships between variables.} \tn 
% Row Count 37 (+ 12)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Time Series Forecasting Cheat Sheet (cont)}}  \tn
% Row 10
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Machine Learning for Time Series} \tn 
% Row Count 1 (+ 1)
% Row 11
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Regression Models: Linear regression, random forest, support vector machines (SVM), or gradient boosting algorithms can be used with appropriate feature engineering. \{\{nl\}\}\{\{nl\}\}Long Short-Term Memory (LSTM) Networks: A type of recurrent neural network (RNN) suitable for modeling sequential data. \{\{nl\}\}\{\{nl\}\}Convolutional Neural Networks (CNN): Can be applied to time series data by treating the series as an image.} \tn 
% Row Count 10 (+ 9)
% Row 12
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Feature Engineering} \tn 
% Row Count 11 (+ 1)
% Row 13
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Lagged Variables: Include lagged versions of the target variable or other relevant variables as features. \{\{nl\}\}\{\{nl\}\}Rolling Statistics: Compute rolling mean, standard deviation, or other statistics over a window of observations. \{\{nl\}\}\{\{nl\}\}Seasonal Features: Extract features representing day of the week, month, or other seasonal patterns. \{\{nl\}\}\{\{nl\}\}Fourier Transform: Convert time series data to frequency domain to identify periodic components.} \tn 
% Row Count 21 (+ 10)
% Row 14
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Validation and Evaluation Metrics} \tn 
% Row Count 22 (+ 1)
% Row 15
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Train-Validation-Test Split: Split the time series into training, validation, and test sets. \{\{nl\}\}\{\{nl\}\}Evaluation Metrics: Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), Mean Absolute Percentage Error (MAPE), and symmetric MAPE (sMAPE) are commonly used.} \tn 
% Row Count 28 (+ 6)
% Row 16
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Cross-Validation for Time Series} \tn 
% Row Count 29 (+ 1)
% Row 17
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Time Series Cross-Validation: Use rolling window or expanding window techniques to simulate the real-time forecasting scenario.} \tn 
% Row Count 32 (+ 3)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Time Series Forecasting Cheat Sheet (cont)}}  \tn
% Row 18
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Ensemble Methods} \tn 
% Row Count 1 (+ 1)
% Row 19
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Combine forecasts from multiple models or model configurations to improve accuracy and robustness. \{\{nl\}\}Examples include model averaging, weighted averaging, and stacking} \tn 
% Row Count 5 (+ 4)
% Row 20
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Outliers and Anomalies} \tn 
% Row Count 6 (+ 1)
% Row 21
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Identify and handle outliers and anomalies to prevent their influence on the forecasting process. \{\{nl\}\}\{\{nl\}\}Techniques include moving averages, median filtering, or statistical tests.} \tn 
% Row Count 10 (+ 4)
% Row 22
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Handling Missing Data} \tn 
% Row Count 11 (+ 1)
% Row 23
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Imputation Techniques: Use interpolation, mean imputation, or model-based imputation to fill missing values.} \tn 
% Row Count 14 (+ 3)
\hhline{>{\arrayrulecolor{DarkBackground}}--}
\end{tabularx}
\par\addvspace{1.3em}

\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Hyperparameter Tuning Cheat Sheet}}  \tn
% Row 0
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{What are Hyperparameters?} \tn 
% Row Count 1 (+ 1)
% Row 1
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Hyperparameters are configuration settings that are not learned from the data but are set before the training process. \{\{nl\}\}\{\{nl\}\}They control the behavior and performance of machine learning models.} \tn 
% Row Count 5 (+ 4)
% Row 2
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Hyperparameter Tuning Techniques:} \tn 
% Row Count 6 (+ 1)
% Row 3
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Grid Search: Exhaustively searches all possible combinations of hyperparameters within predefined ranges. \{\{nl\}\}\{\{nl\}\}Random Search: Randomly samples hyperparameters from predefined ranges, allowing more efficient exploration. \{\{nl\}\}\{\{nl\}\}Bayesian Optimization: Uses prior knowledge and statistical methods to intelligently search the hyperparameter space. \{\{nl\}\}\{\{nl\}\}Genetic Algorithms: Mimics natural selection to evolve a population of hyperparameter configurations over multiple iterations. \{\{nl\}\}\{\{nl\}\}Automated Hyperparameter Tuning Libraries: Tools like Optuna, Hyperopt, or scikit-learn's GridSearchCV and RandomizedSearchCV can automate the hyperparameter tuning process.} \tn 
% Row Count 20 (+ 14)
% Row 4
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Hyperparameters to Consider} \tn 
% Row Count 21 (+ 1)
% Row 5
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Learning Rate: Controls the step size during model training. \{\{nl\}\}\{\{nl\}\}Number of Hidden Units/Layers: Determines the complexity and capacity of neural networks. \{\{nl\}\}\{\{nl\}\}Regularization Parameters: Control the trade-off between model complexity and overfitting. \{\{nl\}\}\{\{nl\}\}Batch Size: Determines the number of samples processed before updating model weights. \{\{nl\}\}\{\{nl\}\}Dropout Rate: Probability of dropping out units during training to prevent overfitting. \{\{nl\}\}\{\{nl\}\}Activation Functions: Choices like sigmoid, tanh, ReLU, or Leaky ReLU impact the model's non-linearity. \{\{nl\}\}\{\{nl\}\}Optimizer: Algorithms like stochastic gradient descent (SGD), Adam, or RMSprop that update model weights during training. \{\{nl\}\}\{\{nl\}\}Number of Trees and Tree Depth: Parameters for ensemble methods like Random Forest or Gradient Boosting models. \{\{nl\}\}\{\{nl\}\}Kernel Type and Parameters: For models like Support Vector Machines (SVM) that use kernel functions.} \tn 
% Row Count 40 (+ 19)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Hyperparameter Tuning Cheat Sheet (cont)}}  \tn
% Row 6
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Define Hyperparameter Ranges} \tn 
% Row Count 1 (+ 1)
% Row 7
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Establish reasonable ranges for each hyperparameter based on prior knowledge, literature, or experimentation. \{\{nl\}\}\{\{nl\}\}Consider the scale and distribution of values (linear, logarithmic) that make sense for each hyperparameter.} \tn 
% Row Count 6 (+ 5)
% Row 8
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Sequential vs. Parallel Tuning} \tn 
% Row Count 7 (+ 1)
% Row 9
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Sequential tuning explores hyperparameter combinations one by one, allowing feedback from each trial to inform the next. \{\{nl\}\}\{\{nl\}\}Parallel tuning performs multiple hyperparameter evaluations simultaneously, making efficient use of computational resources.} \tn 
% Row Count 13 (+ 6)
% Row 10
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Evaluate and Compare Models} \tn 
% Row Count 14 (+ 1)
% Row 11
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Define an evaluation metric (e.g., accuracy, F1-score, mean squared error) that reflects the performance of interest. \{\{nl\}\}\{\{nl\}\}Keep a record of the performance for each hyperparameter configuration to compare the models later.} \tn 
% Row Count 19 (+ 5)
% Row 12
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Cross-Validation} \tn 
% Row Count 20 (+ 1)
% Row 13
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Use techniques like k-fold cross-validation to estimate the generalization performance of different hyperparameter configurations. \{\{nl\}\}\{\{nl\}\}Avoid tuning hyperparameters on the test set to prevent overfitting and biased performance estimation.} \tn 
% Row Count 25 (+ 5)
% Row 14
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Early Stopping} \tn 
% Row Count 26 (+ 1)
% Row 15
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Monitor a validation metric during training and stop the training process early if performance deteriorates consistently. \{\{nl\}\}\{\{nl\}\}Prevents overfitting and saves computational resources.} \tn 
% Row Count 30 (+ 4)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Hyperparameter Tuning Cheat Sheet (cont)}}  \tn
% Row 16
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Feature Selection and Dimensionality Reduction} \tn 
% Row Count 1 (+ 1)
% Row 17
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Consider using techniques like feature selection or dimensionality reduction algorithms (e.g., PCA) as part of hyperparameter tuning. \{\{nl\}\}\{\{nl\}\}They can influence model performance and help improve efficiency.} \tn 
% Row Count 6 (+ 5)
% Row 18
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Domain Knowledge} \tn 
% Row Count 7 (+ 1)
% Row 19
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Leverage domain knowledge to guide the selection of hyperparameters. \{\{nl\}\}\{\{nl\}\}Prior knowledge can help narrow down the search space and focus on hyperparameters likely to have a significant impact.} \tn 
% Row Count 11 (+ 4)
% Row 20
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Regularize Hyperparameters} \tn 
% Row Count 12 (+ 1)
% Row 21
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Apply regularization techniques like L1 or L2 regularization to hyperparameters. \{\{nl\}\}\{\{nl\}\}Regularization helps control the complexity and prevent overfitting of the models.} \tn 
% Row Count 16 (+ 4)
% Row 22
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Documentation and Reproducibility} \tn 
% Row Count 17 (+ 1)
% Row 23
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Keep a record of the hyperparameter configurations, evaluation metrics, and other relevant details for reproducibility. \{\{nl\}\}\{\{nl\}\}Document the lessons learned and insights gained during the hyperparameter tuning process.} \tn 
% Row Count 22 (+ 5)
\hhline{>{\arrayrulecolor{DarkBackground}}--}
\end{tabularx}
\par\addvspace{1.3em}

\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Model Evaluation and Metrics Cheat Sheet}}  \tn
% Row 0
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Confusion Matrix} \tn 
% Row Count 1 (+ 1)
% Row 1
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{A table that summarizes the performance of a classification model. \{\{nl\}\}\{\{nl\}\}It shows the counts of true positives, true negatives, false positives, and false negatives.} \tn 
% Row Count 5 (+ 4)
% Row 2
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Accuracy} \tn 
% Row Count 6 (+ 1)
% Row 3
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{The proportion of correct predictions over the total number of predictions. \{\{nl\}\}\{\{nl\}\}Accuracy = (TP + TN) / (TP + TN + FP + FN)} \tn 
% Row Count 9 (+ 3)
% Row 4
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Precision} \tn 
% Row Count 10 (+ 1)
% Row 5
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{The proportion of true positive predictions over the total number of positive predictions. \{\{nl\}\}\{\{nl\}\}Precision = TP / (TP + FP)} \tn 
% Row Count 13 (+ 3)
% Row 6
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Recall (Sensitivity or True Positive Rate)} \tn 
% Row Count 14 (+ 1)
% Row 7
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{The proportion of true positive predictions over the total number of actual positives. \{\{nl\}\}\{\{nl\}\}Recall = TP / (TP + FN)} \tn 
% Row Count 17 (+ 3)
% Row 8
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Specificity (True Negative Rate)} \tn 
% Row Count 18 (+ 1)
% Row 9
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{The proportion of true negative predictions over the total number of actual negatives. \{\{nl\}\}\{\{nl\}\}Specificity = TN / (TN + FP)} \tn 
% Row Count 21 (+ 3)
% Row 10
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{F1-Score} \tn 
% Row Count 22 (+ 1)
% Row 11
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{The harmonic mean of precision and recall. \{\{nl\}\}\{\{nl\}\}F1-Score = 2 {\emph{ (Precision }} Recall) / (Precision + Recall)} \tn 
% Row Count 25 (+ 3)
% Row 12
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Receiver Operating Characteristic (ROC) Curve} \tn 
% Row Count 26 (+ 1)
% Row 13
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{A plot of the true positive rate (sensitivity) against the false positive rate (1 - specificity) at various classification thresholds. \{\{nl\}\}\{\{nl\}\}It illustrates the trade-off between sensitivity and specificity.} \tn 
% Row Count 31 (+ 5)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Model Evaluation and Metrics Cheat Sheet (cont)}}  \tn
% Row 14
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Area Under the ROC Curve (AUC-ROC)} \tn 
% Row Count 1 (+ 1)
% Row 15
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{A measure of the overall performance of a binary classification model. \{\{nl\}\}\{\{nl\}\}AUC-ROC ranges from 0 to 1, with higher values indicating better performance.} \tn 
% Row Count 5 (+ 4)
% Row 16
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Mean Squared Error (MSE)} \tn 
% Row Count 6 (+ 1)
% Row 17
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{The average of the squared differences between predicted and actual values. \{\{nl\}\}\{\{nl\}\}MSE = (1/n) * Σ(y\_pred - y\_actual)\textasciicircum{}2} \tn 
% Row Count 9 (+ 3)
% Row 18
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Root Mean Squared Error (RMSE)} \tn 
% Row Count 10 (+ 1)
% Row 19
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{The square root of the mean squared error. \{\{nl\}\}\{\{nl\}\}RMSE = √(MSE)} \tn 
% Row Count 12 (+ 2)
% Row 20
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Mean Absolute Error (MAE)} \tn 
% Row Count 13 (+ 1)
% Row 21
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{The average of the absolute differences between predicted and actual values. \{\{nl\}\}\{\{nl\}\}MAE = (1/n) * Σ|y\_pred - y\_actual|} \tn 
% Row Count 16 (+ 3)
% Row 22
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{R-squared (Coefficient of Determination)} \tn 
% Row Count 17 (+ 1)
% Row 23
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{A measure of how well the regression model fits the data. \{\{nl\}\}\{\{nl\}\}R-squared ranges from 0 to 1, with higher values indicating a better fit.} \tn 
% Row Count 20 (+ 3)
% Row 24
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Mean Average Percentage Error (MAPE)} \tn 
% Row Count 21 (+ 1)
% Row 25
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{The average percentage difference between predicted and actual values. \{\{nl\}\}\{\{nl\}\}MAPE = (1/n) {\emph{ Σ(|y\_pred - y\_actual| / y\_actual) }} 100} \tn 
% Row Count 24 (+ 3)
% Row 26
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Cross-Validation} \tn 
% Row Count 25 (+ 1)
% Row 27
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{A technique to assess the performance of a model on unseen data by splitting the data into multiple folds. \{\{nl\}\}It helps estimate the model's generalization performance and mitigate issues like overfitting.} \tn 
% Row Count 30 (+ 5)
\end{tabularx}
\par\addvspace{1.3em}

\vfill
\columnbreak
\begin{tabularx}{5.377cm}{p{0.4977 cm} p{0.4977 cm} }
\SetRowColor{DarkBackground}
\mymulticolumn{2}{x{5.377cm}}{\bf\textcolor{white}{Model Evaluation and Metrics Cheat Sheet (cont)}}  \tn
% Row 28
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Bias-Variance Trade-off} \tn 
% Row Count 1 (+ 1)
% Row 29
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Bias refers to the error introduced by approximating a real-world problem with a simplified model. \{\{nl\}\}\{\{nl\}\}Variance refers to the model's sensitivity to fluctuations in the training data. \{\{nl\}\}\{\{nl\}\}Balancing bias and variance is crucial for building models that generalize well.} \tn 
% Row Count 7 (+ 6)
% Row 30
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Overfitting and Underfitting} \tn 
% Row Count 8 (+ 1)
% Row 31
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Overfitting occurs when a model performs well on training data but poorly on unseen data. \{\{nl\}\}\{\{nl\}\}Underfitting occurs when a model is too simple to capture the underlying patterns in the data. \{\{nl\}\}\{\{nl\}\}Regularization techniques and proper model complexity selection can help address these issues.} \tn 
% Row Count 15 (+ 7)
% Row 32
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Feature Importance} \tn 
% Row Count 16 (+ 1)
% Row 33
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Techniques like feature importance scores, permutation importance, or SHAP values help identify the most influential features in a model.} \tn 
% Row Count 19 (+ 3)
% Row 34
\SetRowColor{LightBackground}
\mymulticolumn{2}{x{5.377cm}}{Model Selection} \tn 
% Row Count 20 (+ 1)
% Row 35
\SetRowColor{white}
\mymulticolumn{2}{x{5.377cm}}{Compare and select models based on evaluation metrics, cross-validation results, and domain-specific considerations. \{\{nl\}\}\{\{nl\}\}Avoid selecting models solely based on a single metric without considering the context.} \tn 
% Row Count 25 (+ 5)
\hhline{>{\arrayrulecolor{DarkBackground}}--}
\end{tabularx}
\par\addvspace{1.3em}


% That's all folks
\end{multicols*}

\end{document}