Adding sparse matrix from CountVectorizer into dataframe with complimentary information for classifier - keep it in sparse format

you dont need to use a data frame.

convert the numerical features from dataframe to a numpy array:

num_feats = df[[cols]].values

from scipy import sparse

training_data = sparse.hstack((count_vectorizer_features, num_feats))

then you can use a scikit-learn algorithm which supports sparse data.

for GBM, you can use xgboost which supports sparse.


As @AbhishekThakur has already said, you don't have to put your one-hot-encoded data into the DataFrame.

But if you want to do so, you can add Pandas.SparseSeries as a columns:

#vecotrizing text
tf_vectorizer = CountVectorizer(max_df=0.5, min_df=2,
                                max_features=n_features,
                                stop_words='english')

#getting the TF matrix
tf = tf_vectorizer.fit_transform(df.pop('reviewText'))

# adding "features" columns as SparseSeries
for i, col in enumerate(tf_vectorizer.get_feature_names()):
    df[col] = pd.SparseSeries(tf[:, i].toarray().ravel(), fill_value=0)

Result:

In [107]: df.head(3)
Out[107]:
        asin  price      reviewerID  LenReview                  Summary  LenSummary  overall  helpful  reviewSentiment         0  \
0  151972036   8.48  A14NU55NQZXML2        199  really a difficult read          23        3        2          -0.7203  0.002632
1  151972036   8.48  A1CSBLAPMYV8Y0         77                      wha           3        4        0          -0.1260  0.005556
2  151972036   8.48  A1DDECXCGHDYZK        114       wordy and drags on          18        1        4           0.5707  0.004545

   ...    think  thought  trailers  trying  wanted  words  worth  wouldn  writing  young
0  ...        0        0         0       0       1      0      0       0        0      0
1  ...        0        0         0       1       0      0      0       0        0      0
2  ...        0        0         0       0       1      0      1       0        0      0

[3 rows x 78 columns]

Pay attention at memory usage:

In [108]: df.memory_usage()
Out[108]:
Index               80
asin               112
price              112
reviewerID         112
LenReview          112
Summary            112
LenSummary         112
overall            112
helpful            112
reviewSentiment    112
0                  112
1                  112
2                  112
3                  112
4                  112
5                  112
6                  112
7                  112
8                  112
9                  112
10                 112
11                 112
12                 112
13                 112
14                 112
                  ...
parts               16   # memory used: # of ones multiplied by 8 (np.int64)
peter               16
picked              16
point               16
quick               16
rating              16
reader              16
reading             24
really              24
reviews             16
stars               16
start               16
story               32
tedious             16
things              16
think               16
thought             16
trailers            16
trying              16
wanted              24
words               16
worth               16
wouldn              16
writing             24
young               16
dtype: int64