Introduction
This cheat sheet provides a comparison between basic data processing technique as well as machine learning models in both R and Python. |
Load dataset in R
library(datasets) |
Import packages |
data(iris) |
Load dataset |
head(iris) |
Look up the first 6 rows of the dataset |
summary(iris) |
Get summary statistics of each columns |
names(iris) |
Get the column names |
Data preprocessing in R
scaling = preProcess(data, method = c('center', 'scale')) |
Create scaling based on data |
data_scaled = predict(scaling, data) |
Apply scaling to data |
train_partition = createDataPartition(y, p = 0.8, list = FALSE) |
Balanced splitting based on the outcome ( 80/20 split) |
data_train = data[train_partition,] |
Split data into train and test sets |
data_test = data[-train_partition,] |
Split data into train and test sets |
Supervised learning models in R
model = lm(data, y ~ x) |
Simple linear regression |
model = lm(data, y ~ x1 + x2 + x3) |
Multiple linear regression |
summary(model) |
Print summary statistics from linear model |
predictions = predict(object, newdata) |
Make prediction based on the model object |
model = glm(data, y ~ x1 + x2 + x3, family = 'binomial') |
Logistic regression |
model = svm(data, y ~ x1 + x2 + x3, params) |
Support vector machines (SVM) |
model = rpart(data, y ~ x1 + x2 + x3, params) |
Decision trees |
model = randomForest(data, y ~ x1 + x2 + x3, params) |
Random forest |
data_xgb = xgb.DMatrix(data, label) |
Transform the data into DMatrix format |
model = xgb.train(data_xgb, label, params) |
Gradient boosting models |
predictions = knn(train, test, cl, params) |
k-NN with labels cl and parameters (e.g., number of neighbors) |
Unsupervised learning models
model = kmeans(x, params) |
K-Means clustering |
model = prcomp(x, params) |
Principal components analysis (PCA) |
Model performance in R
RMSE(pred, actual) |
Root mean square error |
R2(pred, actual, form = 'traditional' ) |
Proportion of the variance explained by the model |
mean(actual == pred) |
Accuracy (how accurate positive predictions are) |
confusionMatrix(actual, pred) |
Confusion matrix |
auc(actual, pred) |
Area under the ROC curve |
f1Score(actual, pred) |
Harmonic mean of precision and recall |
Data visualization in R
geom_point(x, y, color, size, fill, alpha) |
Scatter plot |
geom_line(x, y, color, size, fill, alpha, linetype) |
Line plot |
geom_bar(x, y, color, size, fill, alpha) |
Bar chart |
geom_boxplot(x, y, color) |
Box plot |
geom_tile(x, y, color, fill) |
Heatmap |
|
|
Import file in Python
import pandas as pd |
Import package |
df = pd.read_csv() |
Read csv files |
df.head(n) |
Look up the first n rows of the dataset |
df.describe() |
Get summary statistics of each columns |
df.columns |
Get column names |
Data Processing in Python
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=0) |
Split the dataset into training (80%) and test (20%) sets |
scaler = StandardScaler() |
Standardize features by removing the mean and scaling to unit variance |
X_train = scaler.fit_transform(X_train) |
Fit and transform scalar on X_train |
X_test = scaler.transform(X_test) |
Transform X_test |
Supervised learning models in Python
model = LinearRegression() |
Linear regression |
model.fit(X_train, y_train) |
Fit linear model |
model.predict(X_test) |
Predict using the linear model |
LogisticRegression().fit(X_train, y_train) |
Logistic regression |
LinearSVC.fit(X_train, y_train) |
Train primal SVM |
SVC().fit(X_train, y_train) |
Train dual SVM |
DecisionTreeClassifier().fit(X_train, y_train) |
Decision tree classifier |
RandomForestClassifier().fit(X_train, y_train) |
Random forest classifier |
GradientBoostingClassifier().fit(X_train, y_train) |
Gradient boosting for classification |
XGBClassifier().fit(X_train, y_train) |
XGboost classifier |
KNeighborsClassifier().fit(X_train, y_train) |
k-NN |
Unsupervised learning models
KMeans().fit(X) |
K-Means clustering |
PCA().fit(X) |
Principal component analysis (PCA) |
Model performance in Python
metrics.mean_squared_error(y_true, y_pred, squared=False) |
Root mean squared error |
metrics.r2_score(y_true, y_pred) |
Proportion of the variance explained by the model |
metrics.confusion_matrix(y_true, y_pred) |
Confusion matrix |
metrics.accuracy_score(y_true, y_pred) |
Accuracy classification score |
metrics.roc_auc_score() |
Compute ROC-AUC from prediction scores |
f1_score(y_true, y_pred, average='macro') |
Harmonic mean of the precision and recall |
Data visualization in Python
sns.scatterplot(x, y, hue, size) |
Scatter plot |
sns.lineplot(x, y, hue, size) |
Line plot |
sns.barplot(x, y, hue) |
Bar chart |
sns.boxplot(x, y, hue) |
Box plot |
sns.heatmap(data, linecolor, linewidth) |
Heatmap |
|
