Understanding ( Apply,Applymap ) for Dataframe & (Apply, Map) for Series

• The main aim of using the Pandas .apply() method is to speed up operations and avoid the use of loops for iterating over your data

# import pandas and numpy library
import pandas as pd
import numpy as np

# List of Tuples
matrix = [(1, 2, 3),
            (4, 5, 6),
            (7, 8, 9)
            ]

# Create a DataFrame object
df = pd.DataFrame(matrix, columns = list('xyz'),
                            index = list('abc'))
display(df)

## Directly printing to see both the outputs for both the axis
# print(df.apply(lambda x : print(x),axis=0))
# print(df.apply(lambda x : print(x),axis=1))

#### Applying

	x	y	z
a	1	2	3
b	4	5	6
c	7	8	9

#### Lets see by applying 'apply' it into a series.
# df['x'] = df['x'].apply(lambda x : x*x)

def new_func(x,m):
    return x+m

# The additional arguments can be a scalar, but cannot be series.
df['x'] = df['x'].apply(new_func,args=(2,))
## This is not possible.
# df['x'] = df['x'].apply(new_func,args=(df['y'],))

df

	x	y	z
a	3	2	3
b	6	5	6
c	9	8	9

#### Lets see by applying 'applymap' it into a series.
# print(df.applymap(lambda x : x+1))
# Applymap cannot be applied to a series object, but series.map is available!!
# print(df['x'].applymap(lambda x : x+1))

axis = 0
if axis == 1 :
    # axis = 1 means rows : moving down, ie index = [x,y,z]
    sliced_series = df.loc['a']
elif axis == 0:
    # axis = 0 means columns : moving right, ie index = [a,b,c]
    sliced_series=df.loc[:,'x']

# Accessing
# Index of series
print(f"Series Name => {sliced_series.index}")
# Name of the series
print(f"Series Name => {sliced_series.name}")

list(sliced_series.index)
# Index of the series
# df.loc['a']
# Axis =
# x=df.loc[:,'y']
# x.name
# sliced_series.name

Series Name => Index(['a', 'b', 'c'], dtype='object')
Series Name => x

['a', 'b', 'c']

# Objects passed to the function are Series objects whose index is either
# the DataFrame's index (axis=0) or the DataFrame's columns (axis=1).
# TODO : Video Explaining the Meaning of Axis:
# 0 : Rows
# 1 : Columns
# ---------------> Axis = 1

# axis = obj._get_axis_number(axis)
#     klass: type[FrameApply]
#     if axis == 0:
#         klass = FrameRowApply
#     elif axis == 1:
#         klass = FrameColumnApply

# Apply function numpy.square() to
# square the values of two rows

def new_f(l):
    return l*l


# Applying function to column y and z
column_transform_df = df.apply(lambda x: new_f(x,x['x']) if x.name in ['y', 'z'] else x,
                            axis = 1)

## Applying function to full series.
column_transform_df = df.apply(lambda x: new_f(x) if x.name in ['y', 'z'] else x,axis = 0)
row_transform_df = df.apply(lambda x: new_f(x) if x.name in ['b', 'c'] else x,axis = 1)


display(df)
# Output
display(f'Column Transformation, Axis=0')
display(column_transform_df)

display(f'Row Transformation, Axis=1')
display(row_transform_df)

	x	y	z
a	1	2	3
b	4	5	6
c	7	8	9

'Column Transformation, Axis=0'

	x	y	z
a	1	4	9
b	4	25	36
c	7	64	81

'Row Transformation, Axis=1'

	x	y	z
a	1	2	3
b	16	25	36
c	49	64	81

Elementwise Apply function to Dataframe

Custom column wise function to replace columns with applied function.

def applyPhoneFilter(df,phone_cols,country_col):
    '''
    This function will apply the phone filter to the dataframe.
    and return the filtered dataframe.
    country_col is the country column name.Its important.
    '''
    phone_cc = ''
    for phone_col in phone_cols:
        phone_list = []

        for index,row in df.iterrows():
            try:
                _phone = str(PhoneCleanerClass.extractPhone(row[phone_col],phone_cc,str(row[country_col]))[0])
                phone_list.append(_phone)
                # print(_phone)
            except Exception as e:
                # print(e)

                phone_list.append('nan')
        df[phone_col] = phone_list

    return df

previous

Pandas

next

Iteration over Dataframe and Series.