In [ ]:

    

import math
import pandas as pd


    

In [ ]:

    

def entropy(df, attribute):
    """
    Calculates the entropy of a dataframe for the passed attribute.
    :param df: DataFrame
    :attribute: the attribute we want to calculate the entropy on
    returns the entropy of the attribute
    """
    entropy = 0.0
    freq = {}
    
    for ii in range(len(df)):
        row = df.iloc[[ii]]
        attribute_val = row[attribute].values[0]
        if (freq.has_key(attribute_val)):
            freq[attribute_val] += 1.0
        else:
            freq[attribute_val]  = 1.0 
    N = len(df)
    for freqi in freq.values():
        entropy += (-freqi/N) * math.log(freqi/N, 2) 
 
    return entropy


    

In [ ]: