In this article we will discuss different ways to select rows and columns in DataFrame.

DataFrame provides indexing labels loc & iloc for accessing the column and rows. Also, operator [] can be used to select columns. Let’s discuss them one by one,

First create a DataFrame object i.e.

students = [ ('jack', 34, 'Sydeny') ,
             ('Riti', 30, 'Delhi' ) ,
             ('Aadi', 16, 'New York') ]

# Create a DataFrame object
dfObj = pd.DataFrame(students, columns = ['Name' , 'Age', 'City'], index=['a', 'b', 'c'])

Contents of DataFrame object dfObj  are,

Original DataFrame : 
   Name  Age      City
a  jack   34    Sydeny
b  Riti   30     Delhi
c  Aadi   16  New York

DataFrame.loc | Select Column & Rows by Name

DataFrame provides indexing label loc for selecting columns and rows by names i.e.

dataFrame.loc[<ROWS RANGE> , <COLUMNS RANGE>]

It selects the specified columns and rows from the given DataFrame.
ROWS OR COLUMN RANGE can be also be ‘:’  and if given in rows or column Range parameter then the all entries will be included for corresponding row or column.

Let’s see how to use it,

Select a Column by Name in DataFrame using loc[]

As we want selection on column only, it means all rows should be included for selected column i.e.

'''
Selecting a Single Column by Column Names
'''
columnsData = dfObj.loc[ : , 'Age' ]

It will return a Series object with same indexes as DataFrame.

a    34
b    30
c    16
Name: Age, dtype: int64

Select multiple Columns by Name in DataFrame using loc[]

Pass column names as list,

# Select only 2 columns from dataFrame and create a new subset DataFrame
columnsData = dfObj.loc[ : , ['Age', 'Name'] ]

It will return a subset DataFrame with same indexes but selected columns only i.e.

   Age  Name
a   34  jack
b   30  Riti
c   16  Aadi

Select a single row by Index Label in DataFrame using loc[]

Now we will pass argument ‘:’ in Column range of loc, so that all columns should be included. But for Row Indexes we will pass a label only,

rowData = dfObj.loc[ 'b' , : ]

It will return a series object with same indexes equal to DataFrame columns names i.e.

Name     Riti
Age        30
City    Delhi
Name: b, dtype: object

Select multiple rows by Index labels in DataFrame using loc[]

Pass row index labels as list,

rowData = dfObj.loc[ ['c' , 'b'] , : ]

It will return a subset DataFrame with same columns as DataFrame but selected indexes only i.e.

   Name  Age      City
c  Aadi   16  New York
b  Riti   30     Delhi

Only Rows with index label ‘b’ & ‘c’ are in returned DataFrame object.

Select multiple row & columns by Labels in DataFrame using loc[]

To select multiple rows & column, pass lists containing index labels and column names i.e.

subset = dfObj.loc[ ['c' , 'b'] ,['Age', 'Name'] ]

It will return a subset DataFrame with given rows and columns i.e.

   Age  Name
c   16  Aadi
b   30  Riti

Only Rows with index label ‘b’ & ‘c’ and Columns with names ‘Age’, ‘Name’ are in returned DataFrame object.

Instead of passing all the names in index or column list we can pass range also i.e.

subset = dfObj.loc[ 'a' : 'c' ,'Age' : 'City' ]

It will return a subset DataFrame with rows from a to c & columns from Age to City i.e.

   Age      City
a   34    Sydeny
b   30     Delhi
c   16  New York

DataFrame.iloc | Select Column Indexes & Rows Index Positions

DataFrame provides indexing label iloc for accessing the column and rows by index positions i.e.

dataFrame.iloc[<ROWS INDEX RANGE> , <COLUMNS INDEX RANGE>]

It selects the columns and rows from DataFrame by index position specified in range. If ‘:’ is given in rows or column Index Range then all entries will be included for corresponding row or column.
Let’s see how to use it.

Our DataFrame object dfObj is,

   Name  Age      City
a  jack   34    Sydeny
b  Riti   30     Delhi
c  Aadi   16  New York

Select a single column by Index position

Select column at index 2 i.e.

dfObj.iloc[ : , 2 ]

It will return a Series object i.e,

a      Sydeny
b       Delhi
c    New York
Name: City, dtype: object

Select multiple columns by Index range

Select columns in column index range [0 to 2),

dfObj.iloc[: , [0, 2]]

It will return a DataFrame object i.e,

   Name  Age
a  jack   34
b  Riti   30
c  Aadi   16

Select multiple columns by Indexes in a list

Select columns at column index 0 and 2,

dfObj.iloc[: , [0, 2]]

It will return a DataFrame object i.e,

   Name      City
a  jack    Sydeny
b  Riti     Delhi
c  Aadi  New York

Select single row by Index Position

Select row at index 2 i.e.

dfObj.iloc[ 1 , : ]

It will return a Series object i.e,

Name     Riti
Age        30
City    Delhi
Name: b, dtype: object

Select multiple rows by Index range

Select rows in row index range 0 to 2,

dfObj.iloc[ 0:2 , : ]

It will return a DataFrame object i.e,

   Name  Age    City
a  jack   34  Sydeny
b  Riti   30   Delhi

Select multiple rows by Index positions in a list

Select rows in row index range 0 to 2,

dfObj.iloc[ 0:2 , : ]

It will return a DataFrame object i.e,

   Name  Age      City
c  Aadi   16  New York
a  jack   34    Sydeny

Select multiple rows & columns by Index positions

Select rows at row index 0 and 2,

dfObj.iloc[[2 ,0 ] , : ]

It will return a DataFrame object i.e,

   Age      City
a   34    Sydeny
c   16  New York

Select multiple rows & columns by Index positions

Select rows at index 0 & 2 . Also columns at row 1 and 2,

dfObj.iloc[[0 , 2] , [1 , 2] ]

It will return following DataFrame object,

   Age      City
a   34    Sydeny
c   16  New York

Select multiple rows & columns by Indexes in a range

Select rows at index 0 to 2 (2nd index not included) . Also columns at row 0 to 2 (2nd index not included),

dfObj.iloc[ 0 : 2 ,  1 : 2 ]

It will return following DataFrame object,

   Age
a   34
b   30

If we try to select an index out of range then it will IndexError.

Selecting Columns in DataFrame using [] operator

To access a single or multiple columns from DataFrame by name we can use dictionary like notation on DataFrame i.e.

Select a Column by Name

column2 = dfObj['Age']

It will return a Series object with same indexes as dataFrame i.e.

a    34
b    30
c    16
Name: Age, dtype: int64

Select multiple columns by Name

Instead of passing a single name in [] we can pass a list of column names i.e.

column2 = dfObj[ ['Age', 'Name'] ]

It will return a DataFrame object containing only specified columns from given DataFrame object i.e.

   Age  Name
a   34  jack
b   30  Riti
c   16  Aadi

On accessing a column name that doesn’t exists it will throw ‘KeyError‘.

Complete example is as follows,

import pandas as pd

def main():
    
    students = [ ('jack', 34, 'Sydeny') ,
                 ('Riti', 30, 'Delhi' ) ,
                 ('Aadi', 16, 'New York') ]
    
    # Create a DataFrame object
    dfObj = pd.DataFrame(students, columns = ['Name' , 'Age', 'City'], index=['a', 'b', 'c']) 
    
    print("Original DataFrame : ", dfObj, sep="\n")
    
    print("***** Select Columns in DataFrame by [] *********")
  
    '''
    Select a Column by Name using []
    '''
    column2 = dfObj['Age']
    
    print("Select column By Name using [] " , column2 , sep='\n')
    print("Type : " , type(column2))

    '''
    Select  Multiple Column by Name using []
    '''
    column2 = dfObj[ ['Age', 'Name'] ]
    
    print("Select multiple columns By Name using [] " , column2 , sep='\n')
    print("Type : " , type(column2))
       
    
    print("**** Selecting by Column Names & Rows Index Labels Using df.loc ******")
    
    '''
    Selecting a Single Column by Column Names
    '''
    columnsData = dfObj.loc[ : , 'Age' ]
    
    print("Select a column By Name using loc " , columnsData , sep='\n')
    print("Type : " , type(columnsData))
    
    '''
    Selecting multiple Columns by Column Names
    '''
    
    # Select only 2 columns from dataFrame and create a new subset DataFrame
    columnsData = dfObj.loc[ : , ['Age', 'Name'] ]
    
    print("Select multiple columns By Name using loc " , columnsData , sep='\n')
    print("Type : " , type(columnsData))
    
    '''
    Selecting a Single Row by Index label
    '''
    
    rowData = dfObj.loc[ 'b' , : ]
    
    print("Select a Single Row " , rowData , sep='\n')
    print("Type : " , type(rowData))
    
    '''
    Selecting multiple Rows by Index labels
    '''

    rowData = dfObj.loc[ ['c' , 'b'] , : ]
    
    print("Select multiple Rows" , rowData , sep='\n')
    
    '''
    Select both Rows & Columns by Index labels
    '''
    
    subset = dfObj.loc[ ['c' , 'b'] ,['Age', 'Name'] ]
    
    print("Select both columns & Rows" , subset , sep='\n')
    
    subset = dfObj.loc[ 'a' : 'c' ,'Age' : 'City' ]
    
    print("Select both columns & Rows with selection range " , subset , sep='\n')
    
    
    print("**** Selecting by Column Indexes & Rows Index Positions Using df.iloc ******")
    
    '''
    Select a single column by Index Position
    '''
    print(" Select column at index 2 ")
    print( dfObj.iloc[ : , 2 ] )
    
    '''
    Select multiple columns by Index range 
    '''
    print(" Select columns in column index range 0 to 2")
    print(dfObj.iloc[:, 0:2])
    
    '''
    Select multiple columns by Indexes in a list 
    '''
    print(" Select columns at column index  0 and 2")
    print(dfObj.iloc[: , [0, 2]])
    
    
    '''
    Select single row by Index Position
    '''
    print(" Select row at index 2 ")
    print( dfObj.iloc[ 1 , :  ] )
    
    '''
    Select multiple rows by Index range 
    '''
    print(" Select rows in row index range 0 to 2")
    print(dfObj.iloc[ 0:2 , : ])
    
    '''
    Select multiple rows by Index positions in a list 
    '''
    print(" Select rows at row index  0 and 2")
    print(dfObj.iloc[[2 ,0 ] , : ])
    
    
    '''
    Select multiple rows & columns by  Index positions
    '''
    print(" Select rows at index 0 & 2 . Also columns at row  1 and 2")
    print(dfObj.iloc[[0 , 2] , [1 , 2] ])

    '''
    Select multiple rows & columns by  Indexes in a range
    '''
    print(" Select rows at index 0 to 2 (2nd index not included) . Also columns at row  0 to 2 (2nd index not included)")
    print(dfObj.iloc[ 0 : 2 ,  1 : 2 ])
 
       
if __name__ == '__main__':
    main()

Output:

Original DataFrame : 
   Name  Age      City
a  jack   34    Sydeny
b  Riti   30     Delhi
c  Aadi   16  New York
***** Select Columns in DataFrame by [] *********
Select column By Name using [] 
a    34
b    30
c    16
Name: Age, dtype: int64
Type :  <class 'pandas.core.series.Series'>
Select multiple columns By Name using [] 
   Age  Name
a   34  jack
b   30  Riti
c   16  Aadi
Type :  <class 'pandas.core.frame.DataFrame'>
**** Selecting by Column Names & Rows Index Labels Using df.loc ******
Select a column By Name using loc 
a    34
b    30
c    16
Name: Age, dtype: int64
Type :  <class 'pandas.core.series.Series'>
Select multiple columns By Name using loc 
   Age  Name
a   34  jack
b   30  Riti
c   16  Aadi
Type :  <class 'pandas.core.frame.DataFrame'>
Select a Single Row 
Name     Riti
Age        30
City    Delhi
Name: b, dtype: object
Type :  <class 'pandas.core.series.Series'>
Select multiple Rows
   Name  Age      City
c  Aadi   16  New York
b  Riti   30     Delhi
Select both columns & Rows
   Age  Name
c   16  Aadi
b   30  Riti
Select both columns & Rows with selection range 
   Age      City
a   34    Sydeny
b   30     Delhi
c   16  New York
**** Selecting by Column Indexes & Rows Index Positions Using df.iloc ******
 Select column at index 2 
a      Sydeny
b       Delhi
c    New York
Name: City, dtype: object
 Select columns in column index range 0 to 2
   Name  Age
a  jack   34
b  Riti   30
c  Aadi   16
 Select columns at column index  0 and 2
   Name      City
a  jack    Sydeny
b  Riti     Delhi
c  Aadi  New York
 Select row at index 2 
Name     Riti
Age        30
City    Delhi
Name: b, dtype: object
 Select rows in row index range 0 to 2
   Name  Age    City
a  jack   34  Sydeny
b  Riti   30   Delhi
 Select rows at row index  0 and 2
   Name  Age      City
c  Aadi   16  New York
a  jack   34    Sydeny
 Select rows at index 0 & 2 . Also columns at row  1 and 2
   Age      City
a   34    Sydeny
c   16  New York
 Select rows at index 0 to 2 (2nd index not included) . Also columns at row  0 to 2 (2nd index not included)
   Age
a   34
b   30