Convert a feature-list Pandas data frame to a feature-matrix.

feature_matrix.py

import pandas as pd
from sklearn.preprocessing import MultiLabelBinarizer

# Feature-list data frame
df = pd.DataFrame(columns = ["features"], index=['Item 1', 'Item 2'])
df['features'] = [["A", "B"], ["C", "D"]]

# Use scikits-learn to create feature matrix and feature names
mlb = MultiLabelBinarizer()
feature_column_name = 'features'
feature_matrix = mlb.fit_transform(df[feature_column_name])
feature_names = mlb.classes_

# Create feature-martrix data frame
feature_df = pd.DataFrame(data=feature_matrix, columns=feature_names, index=df.index)

The binarizer approach came from a answer to my question posted here, http://stackoverflow.com/questions/34052903/forming-a-sparse-feature-matrix-data-frame-in-pandas/34053068#34053068