Show Menu
Cheatography
  1. Home >
  2. Programming >

Pandas cheat sheet Cheat Sheet (DRAFT) by

The most important Pandas abilities

This is a draft cheat sheet. It is a work in progress and is not finished yet.

To Start

import numpy as np
import pandas as pd

Create

pd.DataFrame(
dict/list, 
index = None,  
columns = None)
create DataFrame from list or dictionary
 
df.ind­ex[­names]
set custom indexes
 
pd.Series(
list/np_array/dict, 
index = None)
create series from the list or np_array or dictionary

Input and Output

pd.rea­d_c­sv(­'name', 
index_col = None)
read csv
 
pd.rea­d_e­xce­l('­name')
read excel
 
df.to_­csv­('n­ame', 
index = False)
save to csv
 
df.to_­exc­el(­'name', 
'sheet_name = 'name', 
index = False)
save to excel

Iteration

for lab, row in df.iterrrows():
 print(lab)
 print(row)

Functi­ons­/Me­thods

s.drop­(ro­w_i­ndex, axis = 0)
drop values from rows of series
 
df.dro­p(c­ol_­name, axis = 0)
drop values from columns
 
df.dro­p(c­olu­mns­=[c­ol_­names])
drop columns from DataFrame
 
df.dro­p_d­upl­ica­tes()
remove duplicate rows (only considers columns)
 
df.sor­t_i­ndex(by = col_names)
sort by the values along an axis
 
df.sort_values(
by = col_names, 
ascending = False)
order rows by values of a column high to low
 
df.ren­ame­(co­lumns = 
{'old_name':'new_name'})
rename the columns of a DataFrame
 
df.rank()
assign ranks to entries
 
pd.con­cat­([d­f1,­df2])
append rows of DataFrames
 
len(df)
number of rows in DataFrame
 
df1.jo­­in­(df2)
join two DataFrames
 
df['co­l_n­ame­'].u­ni­que()
return unique values from column
 
df[col_name].apply(
func/type.method)
apply function to column
 
df.app­­ly­(­f­un­­c/t­­yp­e.m­e­thod)
apply function
 

Extract

df[col­_names]
series of column
 
df[[co­l_n­ames]]
select column
 
df[sta­rt:end]
select many columns
 
df.loc­[in­dex­_name]
select row
 
df.ilo­c[i­nde­x_num]
select rows
 
df.loc­[[i­nde­x_n­ames]]
select rows
 
df.ilo­c[[­ind­ex_­nums]]
select rows
 
df.loc­[[i­nde­x_n­ames], 
[col_names]]
select rows and columns
 
df.loc[:, [col_n­ames]]
select all rows and few columns
 
df.iloc[:, [col_n­ums]]
select all rows and few columns
 
df.head(n)
select first n rows
 
df.tail(n)
select last n rows
 
df.fil­ter­(regex = 'regex')
select columns whose name matches regular expression regex

Boolean Operators

df[np.logica­l_and(
con1, con2, ...)]
1 'and' 2 condition ...
 
df.loc­[con1 & con2]
1 'and' 2 condition ...
 
df[np.logical_or(
con1, con2, ...)]
1 'or' 2 condition ...
 
df.loc­[con1 | con2]
1 'or' 2 condition ...
 
df[np.logica­l_not(con)]
'not' condition
 
df.loc[ ~ con1]
'not' condition
 
df[var]
for condition

Get DataFrame Inform­ation

df.shape
(rows, colums)
 
df.index
decribe index
 
df.columns
describe DataFrame columns
 
df.info()
info on DataFrame
 
df.count()
number of non_NA values for columns
 
df.des­cribe()
summary statics

Math

df.sum()
sum of values for columns
 
df.cum­sum()
cummul­ative sum of values
 
df.min() 
 df.max()
minimum values for columns
maximum values for columns
 
df.med­ian()
median of values columns
 
df.mean()
mean of values for columns
 
df.std()
standard deviation of each object
 
df.var()
variance of each object