Converts a categorical design matrix X to a sparse CSR matrix, then writes to SVM-lite format.
This takes a pandas DataFrame with categorical features, converts category values to a sparse one-hot representation, then writes the sparse matrix to SVM-light format.
SVM-light is a text-based format, with one sample per line. It does not store zero valued features hence is suitable for sparse dataset.
The first element of each line can be used to store a target variable to predict.
X : pandas DataFrame, shape = [n_samples, n_features] Training vectors, where n_samples is the number of samples and n_features is the number of features.
y : array-like, shape = [n_samples] Target values.
filename : string or file-like in binary mode If string, specifies the path that will contain the data. If file-like, data will be written to f. f should be opened in binary mode.
cat_columns: array-like List of categorical columns
num_columns: array-like, optional List of numerical columns
zero_based : boolean, optional Whether column indices should be written zero-based (True) or one-based (False).
comment : string, optional Comment to insert at the top of the file. This should be either a Unicode string, which will be encoded as UTF-8, or an ASCII byte string.
query_id : array-like, shape = [n_samples] Array containing pairwise preference constraints (qid in svmlight format).
import pandas as pd
import numpy as np
category_data_1 = ['tcp','udp','udp','tcp','dns','tcp']
category_data_2 = ['red','blue','red','green','blue','red']
numerical_data = [1,2,1,1,3,4]
data = {'category_data_1': category_data_1,
'category_data_2': category_data_2,
'numerical_data':numerical_data}
X = pd.DataFrame(data)
y = np.array([1.,0.,1.,1.,0.,0.])
cat_columns = ['category_data_1', 'category_data_2']
num_columns = ['numerical_data']
dump_categorical_df_to_svm_light(X, y, 'example', cat_columns, num_columns)
head example
# Generated by dump_svmlight_file from scikit-learn 0.13-git
# Column indices are zero-based
1.000000 2:1.0000000000000000e+00 4:2.0000000000000000e+00
0.000000 0:1.0000000000000000e+00 2:1.0000000000000000e+00 4:2.0000000000000000e+00
1.000000 0:1.0000000000000000e+00 4:2.0000000000000000e+00
1.000000 2:1.0000000000000000e+00 3:1.0000000000000000e+00 4:1.0000000000000000e+00
0.000000 1:1.0000000000000000e+00 2:1.0000000000000000e+00 4:3.0000000000000000e+00
0.000000 2:1.0000000000000000e+00 4:5.0000000000000000e+00