Python Pandas Module Cheatsheet

Import Pandas Module

import pandas as pd

Creating DataFrames

From a Dictionary

df1 = pd.DataFrame({
    'name': ['John Smith', 'Jane Doe', 'Joe Schmo'],
    'address': ['123 Main St.', '456 Maple Ave.', '789 Broadway'],
    'age': [34, 28, 51]
})

Note: All columns need the same number of elements.

From a List

df2 = pd.DataFrame([
    ['John Smith', '123 Main St.', 34],
    ['Jane Doe', '456 Maple Ave.', 28],
    ['Joe Schmo', '789 Broadway', 51]
    ],
    columns=['name', 'address', 'age'])

From a CSV File

df3 = pd.read_csv('sample.csv')

Viewing DataFrames

Show top lines

print(df.head())     # print first 5 lines
print(df.head(10)    # print first 10 lines

Get Informations about DataFrame Data

print(df.info())

Prints Number of records, and name, datatype and number of filled lines of each individual column.

Select one Column

Given a dataframe there are two ways to return a column:

column_way1 = df['columnname']

# this only works if the columnname has no special characters and spaces
column_way2 = df.columnname

The type of the returned value from a selected column is called a Series, if only one column was selected.

print(type(column_way1))

# returns:
# <class 'pandas.core.series.Series'>

Select Multiple Columns

Selecting columns 3 and 7 from a dataframe with multiple columns:

new_df = df[['column3', 'column7']]

Note: Double set of brackets [[]] is mandatory. The type of the returned value from two selected columns is a DataFrame.

print(type(new_df))

# returns: 
# <class 'pandas.core.frame.DataFrame'>

Select one Row

DataFrames are zero-indexed. This returns the third row from the DataFrame:

new_df = df.iloc[2]

Select Multiple Rows

Rows are zero-indexed:

# from 3rd row (zero-indexed) to but not incl. 7th row
new_df = df.iloc[3:7]

# from zeroth row to but not incl. 4th row
new_df = df.iloc[:4]

# from 3rd to last row to the final row
new_df = df.iloc[-3:]

Select Multiple Rows with Logical Statement

Comparing against single values:

new_df = df[df.a_column == 'desired value']
new_df = df[df.a_column != 'desired value']

new_df = df[(df.a_column == 'desired value') |
            (df.a_column == 'another desired value')]

new_df = df[df.b_column > 5]

Select Multiple Rows by List of Values

Comparing against a list of values:

new_df = df[df.a_column.isin(['desired value','another desired value'])]

Resetting Indices

Selecting parts of dataframes leads to non-consecuritve indices. To correct for that:

# Creates new dataframe with consecurive indices and a new column holding old indices
new_df = df.reset_index()

# Creates new dataframe with consecurive indices without a column holding old indices
new_df = df.reset_index(drop=True)

# Replaces indices inplace  without extra column for old indices
df.reset_index(inplace=True, drop=True)

Modifying DataFrames

Adding Columns

Add a new column by use of a list of the same length as the existing dataframe:

df['Quantity'] = [100, 150, 50, 35]

Add a new column that contains the same value in all rows:

df['In Stock'] = True

Adding a Calculated Column

Add a column calculated based on the contents of another column:

df['Margin'] = df.Price - df.Cost

Perform a Function on a Column

df['Name'] = df.Name.apply(str.upper)