one-hot encoding on multi-dimension arrays, using pandas or scikit-learn

Question

I am trying to encode one-hot for my data frame. It is a multi dimension array and I am not sure how to do this. The data frame may look like this:

df = pd.DataFrame({'menu': [['Italian', 'Greek'], ['Japanese'], ['Italian','Greek', 'Japanese']], 'price': ['$$', '$$', '$'], 'location': [['NY', 'CA','MI'], 'CA', ['NY', 'CA','MA']]})

The output I want is something like this:

df2 = pd.DataFrame({'menu': [[1,1,0], [0,0,1], [1,1,1]], 'price': [[1,0], [1,0], [0,1]], 'location': [[1,1,1,0], [0,1,0,0], [1,1,0,1]]})

I am not sure how this can be done using pd.get_dummies or scikit-learn. Can someone help me?

Answer 1

You can use:

#create list with one item values
df = df.applymap(lambda x: x if isinstance(x, list) else [x])
print (df)
       location                        menu price
0  [NY, CA, MI]            [Italian, Greek]  [$$]
1          [CA]                  [Japanese]  [$$]
2  [NY, CA, MA]  [Italian, Greek, Japanese]   [$]

from sklearn.preprocessing import MultiLabelBinarizer

mlb = MultiLabelBinarizer()
#create Series for each column by list comprehension
vals = [pd.Series(mlb.fit_transform(df[x]).tolist()) for x in df.columns]
#concat to df
df2 = pd.concat(vals, keys=df.columns, axis=1)
print (df2)

       location       menu   price
0  [1, 0, 1, 1]  [1, 1, 0]  [0, 1]
1  [1, 0, 0, 0]  [0, 0, 1]  [0, 1]
2  [1, 1, 0, 1]  [1, 1, 1]  [1, 0]