Rolling Window Calculations On A Pandas DataFrame

Rolling window calculations on a pandas DataFrame instance is performed by calling the rolling() method to obtain a Rolling instance and calling a standard method like mean() or calling the apply() method.

Dataframe Computations Updated 27 Jul 2020 1 min read

Overview

  • Rolling window calculations involve taking subsets of data, where subsets are of the same or varying length and performing mathematical calculations on them.
  • The pandas Rolling class supports rolling window calculations on Series and DataFrame classes.
  • Rolling class has the popular math functions like sum(), mean() and other related functions implemented. Through the apply() method custom math operations can be performed on a rolling window.

Example

python
# Example Python program that performs rolling

            # window calculations on a pandas DataFrame

import pandas as pds

metrics = {"Pressure":(90, 92, 91, 95, 94),

                       "FlowRate":(35, 36, 36, 38, 37)};

df = pds.DataFrame(data = metrics,

                               index = ("Sensor 1", "Sensor 2", "Sensor 3",

                                           "Sensor 4", "Sensor 5"));

# Sum every two metrics

            print(df.rolling(window=2).sum());

Output

Output
          Pressure  FlowRate

            Sensor 1       NaN       NaN

            Sensor 2     182.0      71.0

            Sensor 3     183.0      72.0

            Sensor 4     186.0      74.0

            Sensor 5     189.0      75.0