Transform pandas DataFrames using apply() and applymap() functions

A pandas dataframe elements are transformed by invoking the methods apply(), applymaps() which take a function as a parameter that works on each element, each row or column respectively. The Python example applies math.sin() to each of the DataFrame elements and uses np.sum() on each of the rows and columns.

Dataframe Computations Updated 8 Aug 2019 2 min read

Overview

  • A function can be applied on a pandas DataFrame object to transform its elements. The pandas DataFrame is designed in such a way that each call to the applied function, can be passed of data: either row-wise, column-wise or element wise.
  • When the function applied on the DataFrame is designed to take a pandas series**or an ndarray, the method to be used is **DataFrame.apply().
  • On the other hand if the function to be applied on the DataFrame takes only one Python object as a parameter then the method to be used is DataFrame.applymap().

Applying a Python function to each row or column of a pandas DataFrame

  • The example below sums the elements of a pandas DataFrame both row-wise as well column-wise using the numpy.sum() function.
  • The Python code uses the method apply(), which takes a function as parameter and applies the function on each column or each row of the DataFrame. Method apply() works on a pandas series or a ndarray.

Example

python
# Example Python program to apply a function

# that transforms the elements of a pandas DataFrame

import pandas as pd

import math

import numpy as np

numericData = [(2, 4, 6),

               (8, 10, 12),

               (45, 50, 55)];

dataFrame = pd.DataFrame(data=numericData);

sum       = dataFrame.apply(np.sum);

print("DataFrame:");

print(dataFrame);

print("Sum of each DataFrame column:")

print(sum);

print(type(sum));

print("Sum of each DataFrame row:")

sum       = dataFrame.apply(np.sum, axis=1);

print(sum);

print(type(sum));

Output

Output
DataFrame:

    0   1   2

0   2   4   6

1   8  10  12

2  45  50  55

Sum of each DataFrame column:

0    55

1    64

2    73

dtype: int64

<class 'pandas.core.series.Series'>

Sum of each DataFrame row:

0     12

1     30

2    150

dtype: int64

<class 'pandas.core.series.Series'>

Applying a Python function to each element of a pandas DataFrame

  • The example below applies math.sin() function on a DataFrame that has angle data using the method DataFrame.applymap(). For each element of the DataFrame the function math.sin() is called once. The result is returned as a new DataFrame instance that has the transformed elements.

Example

python
# Example Python program to apply a function

            # that transforms the elements of a pandas DataFrame

import pandas as pd

            import math

panelAngles = [(15, 20, 25),

                           (30, 35, 40),

                           (45, 50, 55)];

rows        = ["R1", "R2", "R3"];

            columnNames = ["C1", "C2", "C3"];

dataFrame = pd.DataFrame(data=panelAngles, index = rows, columns = columnNames);

            sinValues = dataFrame.applymap(math.sin);

print("DataFrame:");

            print(dataFrame);

print("The function math.sine applied on the DataFrame elements:")

            print(sinValues);

Output

Output
DataFrame:

    C1  C2  C3

R1  15  20  25

R2  30  35  40

R3  45  50  55

The function math.sine applied on the DataFrame elements:

          C1        C2        C3

R1  0.650288  0.912945 -0.132352

R2 -0.988032 -0.428183  0.745113

R3  0.850904 -0.262375 -0.999755