Source

pysvmlight / examples / simple.py

Full commit
import svmlight
#import nltk
import pickle

training_data = __import__('data').train0
test_data = __import__('data').test0

# train a model based on the data
model = svmlight.learn(training_data, type='classification', verbosity=0)

# model data can be stored in the same format SVM-Light uses, for interoperability
# with the binaries.
svmlight.write_model(model, 'my_model.dat')

#with open("model.pickle", 'wb') as f:
#    pickle.dump(model, f)

# classify the test data. this function returns a list of numbers, which represent
# the classifications.
predictions = svmlight.classify(model, test_data)
for p in predictions:
    print '%.8f' % p