How to implement Expectation Maximization Imputation method using python?

Question

From various resources, I came to know that Imputation using Expectation Maximization method is better than Mean Imputation for imputing missing data. But no source have explained how to implement it in python.

I looked into scikit-learn, fancyimpute packages, but they have not mentioned anything about Expectation Maximization method.

It would be very helpful, if you can provide link to documentation which explain implementation with example, or provide code to implement Expectation Maximization method for missing data.

Answer 1

import impyute as impy

data_missing = pd.DataFrame(np.array([[np.NaN, 0.6, np.NaN], [np.NaN, 0.25, 
0.3], [0.6, 0.7, np.NaN]]), columns=['a', 'b', 'c'])
em_imputed = impy.em(data_missing)

output:
   a     b    c
0  0.6  0.60  0.3
1  0.6  0.25  0.3
2  0.6  0.70  0.3

em function will return dataframe type