def _test():
dset = create_dataset('tests/lenses.mff')
dset.train.normalize_attributes()
for e in dset.train.examples:
print(e)
classifier = NeuralNetwork(trainset=dset.train, max_error=.2, debug=True)
evaluator = Evaluator(classifier)
evaluator.holdout(.5)
dset = create_dataset('tests/lenses.mff')
dset.train.nominal_to_linear()
print(dset)
classifier = NeuralNetwork(trainset=dset.train, debug=True, max_error=.1, j=10)
evaluator = Evaluator(classifier)
evaluator.holdout(.2)
dset = create_dataset('tests/test_data/iris-binary.mff')
classifier = NeuralNetwork(trainset=dset.train, debug=True, max_error=.1)
classifier.train(dset.train)
dset = create_dataset('tests/test_data/votes.mff')
classifier = NeuralNetwork(trainset=dset.train, debug=True, max_error=.1)
classifier.train(dset.train)
dset = create_dataset('tests/test_data/mushroom.mff')
classifier = NeuralNetwork(trainset=dset.train, debug=True)
classifier.train(dset.train)
dset = create_dataset('tests/test_data/soybean.mff')
classifier = NeuralNetwork(trainset=dset.train, debug=True)
classifier.train()
def evaluate(classifier, testset=None, holdout=None, folds=10):
"""Create evaulator object.
Args:
classifier (Classifier): desired classifier to run
testset (DataSet): testset to run classification accuracies/tests
outfile (str): filepath of target output file
holdout (float): desired split for the hold-out method
folds (int): number of folds for cross validation
"""
evaluator = Evaluator(classifier)
if testset:
pass
elif holdout:
evaluator.holdout(holdout)
else: # runing folds
evaluator.cross_validate(folds)
return evaluator
