Rolling window calculations on a pandas series

Computations can be performed on each subset of a pandas Series instance which are called rolling window calculations. The subsets are of uniform length. The class Rolling supports several popular computations like average,standard deviations and others including custom computations.

Series Computations Updated 25 Jul 2020 2 min read

Overview

  • Often it is required to perform calculations on the subsets of data of specific length continuously. For example, intra-day stock traders calculate various technical indicators using the past 14 minutes data continously. These subsets of the data are called as rolling windows.
  • The Rolling class in pandas implements a rolling window for the Series and DataFrame classes. A call to the method rolling() on a series instance returns a Rolling object.
  • A Rolling instance supports several standard computations like average, standard deviation and others. A pandas Rolling instance also supports the apply() method through which a function performing custom computations can be called.

Example 1 - Performing a custom rolling window calculation on a pandas series

python
# Example Python program that peforms

            # rolling computations on a pandas data series

            import pandas as pds

# A Python function definition

            # that decides whether an account should get locked

            def donotlock(collection):

                c1 = collection.astype(bool);

                falseCount = 0;

    for status in c1:

                    if status is False:

                        falseCount = falseCount + 1;

    if falseCount >= 3:

                    return False;

                else:

                    return True;

loginStatus    = [True, True, True, False, True, False, False, False];

sr             = pds.Series(data = loginStatus);

# Define a policy that an account will be locked after

            # three consecutive login failures

            tolerance     = 3;

            window         = sr.rolling(window=tolerance);

print(window.apply(func=donotlock, raw=False).astype(bool));

Output

Output
0     True

            1     True

            2     True

            3     True

            4     True

            5     True

            6     True

            7    False

            dtype: bool

Example 2 - Performing a standard rolling window calculation on a pandas Series

python
# Example Python program that calculates

            # the moving average of a stock price

            # over 30 ticks

            import pandas as pds

# Price data as a Python list

            price = [67.0, 67.1, 67.6, 67.0, 67.75, 68.0,

                     67.8, 68.2, 67.5, 67.6, 67.4, 67.0,

                     66.9, 67.2, 67.5, 67.0, 66.8, 66.7,

                     67.0, 67.25, 67.40, 67.20, 67.30, 67.70,

                     68.1, 68.5, 68.2, 67.75, 67.9, 67.6];

# Create a pandas series

            sr = pds.Series(data = price);

# Calculate the moving average

            tickCount = 14;

            print("Moving averages:");

            print(sr.rolling(window=tickCount).mean());

Output

Output
Moving averages:

            0           NaN

            1           NaN

            2           NaN

            3           NaN

            4           NaN

            5           NaN

            6           NaN

            7           NaN

            8           NaN

            9           NaN

            10          NaN

            11          NaN

            12          NaN

            13    67.432143

            14    67.467857

            15    67.460714

            16    67.403571

            17    67.382143

            18    67.328571

            19    67.275000

            20    67.246429

            21    67.175000

            22    67.160714

            23    67.167857

            24    67.217857

            25    67.325000

            26    67.417857

            27    67.457143

            28    67.485714

            29    67.528571

            dtype: float64