def test_basic():
view = skdata.iris.view.KfoldClassification(4)
# fit a model using the default Perceptron args:
def fit_default():
algo = SklearnClassifier(sklearn.linear_model.Perceptron)
mean_test_error = view.protocol(algo)
assert len(algo.results['loss']) == 4
assert len(algo.results['best_model']) == 4
for loss_report in algo.results['loss']:
print loss_report['task_name'] + \
(": err = %0.3f" % (loss_report['err_rate']))
return mean_test_error
default_test_error = fit_default()
algo = SklearnClassifier(AutoPerceptron)
mean_test_error = view.protocol(algo)
assert len(algo.results['loss']) == 4
assert len(algo.results['best_model']) == 4
for loss_report in algo.results['loss']:
print loss_report['task_name'] + \
(": err = %0.3f" % (loss_report['err_rate']))
print "MEAN OPT ERROR", mean_test_error
print "MEAN DEFAULT ERROR", default_test_error
assert mean_test_error <= default_test_error
assert mean_test_error < 0.15 # 0.147 on Feb 19, 2013
assert default_test_error > 0.3 # 0.305 on Feb 19, 2013
def fit_default():
algo = SklearnClassifier(sklearn.linear_model.Perceptron)
mean_test_error = view.protocol(algo)
assert len(algo.results['loss']) == 4
assert len(algo.results['best_model']) == 4
for loss_report in algo.results['loss']:
print loss_report['task_name'] + \
(": err = %0.3f" % (loss_report['err_rate']))
return mean_test_error
