def test_syntax11_learner(self):
df = pandas.DataFrame(
dict(education=['A', 'B', 'A', 'B', 'A'],
workclass=['X', 'X', 'Y', 'Y', 'Y'],
y=[1, 0, 1, 0, 0]))
X = df.drop('y', axis=1)
exp = Pipeline([
OneHotVectorizer() << {
'edu1': 'education'
},
OneHotHashVectorizer() << {
'edu2': 'education'
},
FastLinearBinaryClassifier(max_iterations=1) << {
'Features': ['edu1', 'edu2'],
Role.Label: 'y'
}
])
exp.fit(df)
prediction = exp.predict(X)
assert isinstance(prediction, pandas.DataFrame)
assert sorted(list(
prediction.columns)) == ['PredictedLabel', 'Probability', 'Score']
assert prediction.shape == (5, 3)
def test_syntax5_regular_expression(self):
# REVIEW: not implemented yet
# The best would be to handle regular expression inside nimbusml.
# It could be handled in entrypoint.py just before calling nimbusml.
# It can be handled inside Pipeline if it is aware of
# the input schema.
df = pandas.DataFrame(
dict(education=['A', 'B', 'A', 'B', 'A'],
workclass=['X', 'X', 'Y', 'Y', 'Y'],
y=[1, 0, 1, 0, 0]))
X = df.drop('y', axis=1)
y = df['y']
exp = Pipeline([
OneHotVectorizer() << {
'f1': 'education'
},
OneHotHashVectorizer() << {
'f2': 'education'
},
OneHotVectorizer(max_num_terms=2) << {
'f3': 'workclass'
},
Concat() << {
'Features': 'f[0-9]+'
},
FastLinearBinaryClassifier(max_iterations=1) << 'Features'
])
exp.fit(X, y)
prediction = exp.predict(X)
assert isinstance(prediction, pandas.DataFrame)
assert sorted(list(
prediction.columns)) == ['PredictedLabel', 'Probability', 'Score']
assert prediction.shape == (5, 3)
def test_syntax6_regular_expression(self):
df = pandas.DataFrame(
dict(education=['A', 'B', 'A', 'B', 'A'],
workclass=['X', 'X', 'Y', 'Y', 'Y'],
y=[1, 0, 1, 0, 0]))
X = df.drop('y', axis=1)
y = df['y']
exp = Pipeline([
OneHotVectorizer() << {
'f1': 'education'
},
OneHotHashVectorizer() << {
'f2': 'education'
},
OneHotVectorizer(max_num_terms=2) << {
'f3': 'workclass'
},
Concat() << {
'Features': ['f%d' % i for i in range(1, 4)]
},
Drop() << '~Features',
FastLinearBinaryClassifier(max_iterations=1)
])
exp.fit(X, y)
prediction = exp.predict(X)
assert isinstance(prediction, pandas.DataFrame)
assert sorted(list(
prediction.columns)) == ['PredictedLabel', 'Probability', 'Score']
assert prediction.shape == (5, 3)
def test_syntax4_fail(self):
df = pandas.DataFrame(
dict(education=['A', 'B', 'A', 'B', 'A'],
workclass=['X', 'X', 'Y', 'Y', 'Y'],
y=[1, 0, 1, 0, 0]))
X = df.drop('y', axis=1)
y = df['y']
exp = Pipeline([
OneHotVectorizer() << {
'edu1': 'education'
},
OneHotHashVectorizer() << {
'edu2': 'education'
},
OneHotVectorizer(max_num_terms=2) << {
'wki': 'workclass'
},
FastLinearBinaryClassifier(max_iterations=1) <<
['edu1', 'edu2', 'wki']
])
try:
exp.fit(X, y)
assert False
except RuntimeError as e:
assert "ConcatTransform() << {'Input': ['edu1', 'edu2', 'wki']}" \
in str(e)
def test_syntax4_fail2(self):
df = pandas.DataFrame(
dict(education=['A', 'B', 'A', 'B', 'A'],
workclass=['X', 'X', 'Y', 'Y', 'Y'],
y=[1, 0, 1, 0, 0]))
X = df.drop('y', axis=1)
y = df['y']
exp = Pipeline([
OneHotVectorizer() << {
'edu1': 'education'
},
OneHotHashVectorizer() << {
'edu2': 'education'
},
OneHotVectorizer(max_num_terms=2) << {
'wki': 'workclass'
},
FastLinearBinaryClassifier(max_iterations=1) <<
['edu1', 'edu4', 'wki']
])
try:
exp.fit(X, y)
raise AssertionError("The test should not reach this line.")
except Exception as e:
assert "Feature column 'edu4' not found" in str(e)
def test_syntax4_dict(self):
df = pandas.DataFrame(
dict(education=['A', 'B', 'A', 'B', 'A'],
workclass=['X', 'X', 'Y', 'Y', 'Y'],
y=[1, 0, 1, 0, 0]))
X = df.drop('y', axis=1)
y = df['y']
exp = Pipeline([
OneHotVectorizer() << {
'edu1': 'education'
},
OneHotHashVectorizer() << {
'edu2': 'education'
},
OneHotVectorizer(max_num_terms=2) << {
'wki': 'workclass'
},
Concat() << {
'Inputs': ['edu1', 'edu2', 'wki']
},
FastLinearBinaryClassifier(max_iterations=1) << 'Inputs'
])
exp.fit(X, y)
prediction = exp.predict(X)
assert isinstance(prediction, pandas.DataFrame)
assert sorted(list(
prediction.columns)) == ['PredictedLabel', 'Probability', 'Score']
assert prediction.shape == (5, 3)
def test_syntax6_change_role(self):
# REVIEW: the pipeline drops all columns but one -->
# nimbusml still thinks the Features are eduction, workclass
# and does not automatically detects that the only remaining
# columns should play that role
# (maybe because the label column is here too even though
# the only remaining column without a role is Features).
df = pandas.DataFrame(dict(education=['A', 'B', 'A', 'B', 'A'],
workclass=['X', 'X', 'Y', 'Y', 'Y'],
y=[1, 0, 1, 0, 0]))
X = df.drop('y', axis=1)
y = df['y']
exp = Pipeline([
OneHotVectorizer() << {'f1': 'education'},
OneHotHashVectorizer() << {'f2': 'education'},
OneHotVectorizer(max_num_terms=2) << {'f3': 'workclass'},
Concat() << {'Features': ['f%d' % i for i in range(1, 4)]},
Drop() << ['education', 'workclass', 'f1', 'f2', 'f3'],
FastLinearBinaryClassifier(maximum_number_of_iterations=1) << ['Features']
])
exp.fit(X, y, verbose=0)
prediction = exp.predict(X)
assert isinstance(prediction, pandas.DataFrame)
assert sorted(list(prediction.columns)) == [
'PredictedLabel', 'Probability', 'Score']
assert prediction.shape == (5, 3)
def test_ensemble_supports_user_defined_transforms(self):
test2_df = test_df.copy(deep=True)
test2_df = test2_df.append(pd.DataFrame({'c1': [9, 11], 'c2': [1, 1]}))
r1 = OrdinaryLeastSquaresRegressor(**olsrArgs)
r1.fit(train_df)
result1 = r1.predict(test2_df)
r2 = OnlineGradientDescentRegressor(**ogdArgs)
r2.fit(train_df)
result2 = r2.predict(test2_df)
r3 = LightGbmRegressor(**lgbmArgs)
r3.fit(train_df)
result3 = r3.predict(test2_df)
r1 = OrdinaryLeastSquaresRegressor(**olsrArgs)
r2 = OnlineGradientDescentRegressor(**ogdArgs)
r3 = LightGbmRegressor(**lgbmArgs)
pipeline = Pipeline([
RangeFilter(min=0, max=10, columns='c1'),
VotingRegressor(estimators=[r1, r2, r3], combiner='Average')
])
pipeline.fit(train_df)
result4 = pipeline.predict(test2_df)
self.assertEqual(len(result4), 3)
average1 = (result1[0] + result2[0] + result3[0]) / 3
average2 = (result1[1] + result2[1] + result3[1]) / 3
average3 = (result1[2] + result2[2] + result3[2]) / 3
self.assertAlmostEqual(average1, result4.loc[0, 'Score'], places=5)
self.assertAlmostEqual(average2, result4.loc[1, 'Score'], places=5)
self.assertAlmostEqual(average3, result4.loc[2, 'Score'], places=5)
def test_syntax3(self):
df = pandas.DataFrame(
dict(education=['A', 'B', 'A', 'B', 'A'],
workclass=['X', 'X', 'Y', 'Y', 'Y'],
y=[1, 0, 1, 0, 0]))
X = df.drop('y', axis=1)
y = df['y']
exp = Pipeline([
OneHotVectorizer() << {
'edu1': 'education'
},
OneHotHashVectorizer() << 'education',
OneHotVectorizer(max_num_terms=2) << 'workclass',
# Currently the learner does not use edu1
# unless it is specified explicitely so nimbusml
# does not do what the syntax implicetely tells.
# We need to modify either the bridge to look into
# every available column at one step.
FastLinearBinaryClassifier(max_iterations=1)
])
exp.fit(X, y)
prediction = exp.predict(X)
assert isinstance(prediction, pandas.DataFrame)
assert sorted(list(
prediction.columns)) == ['PredictedLabel', 'Probability', 'Score']
assert prediction.shape == (5, 3)
def test_syntax10_weights(self):
df = pandas.DataFrame(dict(education=['A', 'B', 'A', 'B', 'A'],
workclass=['X', 'X', 'Y', 'Y', 'Y'],
weight=[1., 1., 1., 2., 1.],
y=[1.1, 2.2, 1.24, 3.4, 3.4]))
X = df.drop(['y', 'weight'], axis=1)
y = df['y']
w = df['weight']
exp = Pipeline([
OneHotVectorizer() << ['workclass', 'education'],
FastLinearRegressor()
])
exp.fit(X, y, weight=w, verbose=0)
assert exp.nodes[-1].feature_column == 'Features'
assert exp.nodes[-1].label_column == 'y'
assert exp.nodes[-1].weight_column == 'weight'
X['weight'] = -5
prediction = exp.predict(X)
assert isinstance(prediction, pandas.DataFrame)
assert list(prediction.columns) == ['Score']
assert prediction.shape == (5, 1)
if prediction['Score'].min() < 1.:
raise Exception(prediction)
if prediction['Score'].max() > 3.6:
raise Exception(prediction)
if len(set(prediction['Score'])) < 4:
raise Exception(prediction)
def test_test(self):
transformed_data, transformed_data_df = transform_data()
fl = FastLinearRegressor(
feature=[
'parity',
'in',
'sp',
'stratum'],
label='age')
flpipe = Pipeline([fl])
flpipe.fit(transformed_data)
metrics, scores = flpipe.test(transformed_data, output_scores=True)
metrics_df, scores_df = flpipe.test(
transformed_data_df, output_scores=True)
assert_array_equal(scores, scores_df)
assert_array_equal(metrics, metrics_df)
flpipe.fit(
transformed_data_df.drop(
'age',
axis=1),
transformed_data_df['age'])
metrics, scores = flpipe.test(transformed_data, output_scores=True)
metrics_df, scores_df = flpipe.test(
transformed_data_df, output_scores=True)
assert_array_equal(scores, scores_df)
assert_array_equal(metrics, metrics_df)
def test_metrics_evaluate_clusterer(self):
np.random.seed(0)
df = get_dataset("iris").as_df()
df.drop(['Species'], inplace=True, axis=1)
df.Label = [1 if x == 1 else 0 for x in df.Label]
X_train, X_test, y_train, y_test = \
train_test_split(df.loc[:, df.columns != 'Label'], df['Label'])
lr = KMeansPlusPlus(n_clusters=2, initialization_algorithm="Random")
e = Pipeline([lr])
e.fit(X_train, y_train.to_frame(), verbose=0)
metrics, _ = e.test(X_test, y_test)
# if abs(metrics['NMI'][0] - 0.7) >= 0.15:
# raise AssertionError("NMI loss should be %f not %f" % \
# (0.7, metrics['NMI'][0]))
# if abs(metrics['AvgMinScore'][0] - 0.014) >= 0.015:
# raise AssertionError("AvgMinScore should be %f not %f" % (\
# 0.014, metrics['AvgMinScore'][0]))
assert_almost_equal(metrics['NMI'][0],
0.7,
decimal=0,
err_msg="NMI loss should be %s" % 0.7)
assert_almost_equal(metrics['AvgMinScore'][0],
0.032,
decimal=2,
err_msg="AvgMinScore should be %s" % 0.014)
def test_pipeline_loaded_from_zip_has_feature_contributions(self):
features = ['age', 'education-num', 'hours-per-week']
model_nimbusml = Pipeline(
steps=[FastLinearBinaryClassifier(feature=features)])
model_nimbusml.fit(train, label)
fc = model_nimbusml.get_feature_contributions(test)
# Save the model to zip
model_filename = get_temp_file(suffix='.zip')
model_nimbusml.save_model(model_filename)
# Load the model from zip
model_nimbusml_zip = Pipeline()
model_nimbusml_zip.load_model(model_filename)
fc_zip = model_nimbusml_zip.get_feature_contributions(test)
assert ['FeatureContributions.' + feature in fc_zip.columns
for feature in features]
assert [fc['FeatureContributions.' + feature].equals(
fc_zip['FeatureContributions.' + feature])
for feature in features]
os.remove(model_filename)
def test_pipeline_with_no_columns_raise(self):
trainData = pd.DataFrame({
"Sentiment": [0, 1, 1, 0, 1, 1],
"SentimentText": [
"this is train ", "review ", "sentence ", "an apple",
"sentence 22", "another one one one"
]
})
ppl = Pipeline([
NGramFeaturizer(word_feature_extractor=n_gram()),
LightGbmClassifier()
])
assert ppl is not None
# Bug 147697
info = ppl.get_fit_info(trainData[["SentimentText"]],
trainData["Sentiment"])
assert len(info) == 2
assert len(info[0]) == 3
with self.assertRaises(RuntimeError):
# Message
# System.InvalidOperationException:
# 'LightGBM Error, code is -1, error message is
# 'Cannot construct Dataset since there are not useful features.
# It should be at least two unique rows.
# If the num_row (num_data) is small,
# you can set min_data=1 and min_data_in_bin=1 to fix this.
# Otherwise please make sure you are using the right dataset.'
ppl.fit(trainData[["SentimentText"]], trainData["Sentiment"])
def test_model_summary(self):
for learner in learners:
pipeline = Pipeline(
[OneHotVectorizer() << categorical_columns, learner])
train_stream = FileDataStream(train_file, schema=file_schema)
pipeline.fit(train_stream, label_column)
pipeline.summary()
def test_two_pipelines_created_using_idv_binary_data_can_be_combined_in_to_one_model(
self):
"""
This test verifies that two models can be combined
even if the transform increases the number of columns.
"""
# Create and fit a OneHotVectorizer transform using the
# training data and use it to transform the training data.
transform_pipeline = Pipeline([OneHotVectorizer() << 'c0'],
random_state=seed)
transform_pipeline.fit(train_df)
df = transform_pipeline.transform(train_df, as_binary_data_stream=True)
# Create and fit an OnlineGradientDescentRegressor using
# the transformed training data from the previous step.
predictor_pipeline = Pipeline(
[OnlineGradientDescentRegressor(label='c2', feature=['c0', 'c1'])],
random_state=seed)
predictor_pipeline.fit(df)
# Perform a prediction given the test data using
# the transform and predictor defined previously.
df = transform_pipeline.transform(test_df, as_binary_data_stream=True)
result_1 = predictor_pipeline.predict(df)
# Combine the above Pipelines in to one Pipeline and use
# the new Pipeline to get predictions given the test data.
combined_pipeline = Pipeline.combine_models(transform_pipeline,
predictor_pipeline)
result_2 = combined_pipeline.predict(test_df)
# Verify that the prediction from the combined Pipeline
# matches the prediction from the original two Pipelines.
self.assertEqual(result_1.loc[0, 'Score'], result_2.loc[0, 'Score'])
self.assertEqual(result_1.loc[1, 'Score'], result_2.loc[1, 'Score'])
def test_experiment_loadsavemodel(self):
(train, label) = get_X_y(train_file, label_column, sep=',')
(test, label1) = get_X_y(test_file, label_column, sep=',')
cat = OneHotVectorizer() << categorical_columns
ftree = FastTreesBinaryClassifier()
pipeline = Pipeline([cat, ftree])
pipeline.fit(train, label)
metrics1, scores1 = pipeline.test(test,
label1,
'binary',
output_scores=True)
sum1 = metrics1.sum().sum()
(fd, modelfilename) = tempfile.mkstemp(suffix='.model.bin')
fl = os.fdopen(fd, 'w')
fl.close()
pipeline.save_model(modelfilename)
pipeline2 = Pipeline()
pipeline2.load_model(modelfilename)
metrics2, scores2 = pipeline2.test(test,
label1,
'binary',
output_scores=True)
sum2 = metrics2.sum().sum()
assert_equal(sum1, sum2,
"model metrics don't match after loading model")
def test_notvectorized_output_predictor_model(self):
"""
This test verifies that outputted predictor model from
combined (with featurizers) pipeline runs successfully
on featurized data with no vectors.
"""
df = train_df.drop(['c0'], axis=1)
# Create and fit a RangeFilter transform using the training
# data and use it to transform the training data.
transform_pipeline = Pipeline([RangeFilter(min=0.0, max=4.5) << 'c2'],
random_state=seed)
transform_pipeline.fit(df)
df1 = transform_pipeline.transform(df)
# Create and fit a combined model and spit out predictor model
combined_pipeline = Pipeline([
RangeFilter(min=0.0, max=4.5) << 'c2',
OnlineGradientDescentRegressor(label='c2')
],
random_state=seed)
combined_pipeline.fit(df, output_predictor_model=True)
result_1 = combined_pipeline.predict(df)
# Load predictor pipeline and score featurized data
predictor_pipeline = Pipeline()
predictor_pipeline.load_model(combined_pipeline.predictor_model)
result_2 = predictor_pipeline.predict(df1)
self.assertEqual(result_1.loc[0, 'Score'], result_2.loc[0, 'Score'])
self.assertEqual(result_1.loc[1, 'Score'], result_2.loc[1, 'Score'])
def test_model_summary_not_supported(self):
for learner in learners_not_supported:
pipeline = Pipeline(
[OneHotVectorizer() << categorical_columns, learner])
train_stream = FileDataStream(train_file, schema=file_schema)
pipeline.fit(train_stream, label_column)
assert_raises(TypeError, pipeline.summary)
def test_unpickled_pipeline_has_feature_contributions(self):
features = ['age', 'education-num', 'hours-per-week']
model_nimbusml = Pipeline(
steps=[FastLinearBinaryClassifier(feature=features)])
model_nimbusml.fit(train, label)
fc = model_nimbusml.get_feature_contributions(test)
# Save with pickle
pickle_filename = get_temp_file(suffix='.p')
with open(pickle_filename, 'wb') as f:
pickle.dump(model_nimbusml, f)
# Unpickle model
with open(pickle_filename, "rb") as f:
model_nimbusml_pickle = pickle.load(f)
fc_pickle = model_nimbusml_pickle.get_feature_contributions(test)
assert ['FeatureContributions.' + feature in fc_pickle.columns
for feature in features]
assert [fc['FeatureContributions.' + feature].equals(
fc_pickle['FeatureContributions.' + feature])
for feature in features]
os.remove(pickle_filename)
def test_metrics_evaluate_regressor(self):
np.random.seed(0)
df = get_dataset("iris").as_df()
df.drop(['Species'], inplace=True, axis=1)
df.Label = [1 if x == 1 else 0 for x in df.Label]
X_train, X_test, y_train, y_test = \
train_test_split(df.loc[:, df.columns != 'Label'], df['Label'])
lr = FastTreesRegressor()
e = Pipeline([lr])
e.fit(X_train, y_train.to_frame(), verbose=0)
metrics, _ = e.test(X_test, y_test)
# TODO: debug flucations, and increase decimal precision on checks
assert_almost_equal(metrics['L1(avg)'][0],
0.107,
decimal=1,
err_msg="L1 loss should be %s" % 0.107)
assert_almost_equal(metrics['L2(avg)'][0],
0.0453,
decimal=1,
err_msg="L2(avg) should be %s" % 0.0453)
assert_almost_equal(metrics['Loss-fn(avg)'][0],
0.0453,
decimal=1,
err_msg="Loss-fn(avg)loss should be %s" % 0.0453)
def test_syntax8_label(self):
df = pandas.DataFrame(dict(education=['A', 'B', 'A', 'B', 'A'],
workclass=['X', 'X', 'Y', 'Y', 'Y'],
yy=[1.1, 2.2, 1.24, 3.4, 3.4]))
X = df.drop('yy', axis=1)
exp = Pipeline([
MeanVarianceScaler() << {'new_y': 'yy'},
OneHotVectorizer() << ['workclass', 'education'],
Drop() << 'yy',
FastLinearRegressor() << {'Feature': ['workclass', 'education'],
Role.Label: 'new_y'}
])
exp.fit(df, verbose=0)
assert exp.nodes[-1].feature_column_ == 'Features'
assert exp.nodes[-1].label_column_ == 'new_y'
# The pipeline requires it now as it is transformed all along.
X['yy'] = 0.0
prediction = exp.predict(X, verbose=0)
assert isinstance(prediction, pandas.DataFrame)
assert list(prediction.columns) == ['Score']
assert prediction.shape == (5, 1)
if prediction['Score'].min() < 0.4:
raise Exception(prediction)
if prediction['Score'].max() > 2.00:
raise Exception(prediction)
def test_syntax7_rename(self):
# Error message are usually not informative enough.
# Missing column --> no indication of other columns.
# Error is (one transform should handle it)
# 'The label column 'y' of the training data has a data type
# not suitable for binary classification: Vec<Key<U4, 0-1>, 2>.
# Type must be Bool, R4, R8 or Key with two classes.
df = pandas.DataFrame(
dict(
education=[
'A', 'B', 'A', 'B', 'A'], workclass=[
'X', 'X', 'Y', 'Y', 'Y'], y=[
'red', 'white', 'red', 'white', 'white']))
X = df.drop('y', axis=1)
y = df['y']
exp = Pipeline([
OneHotVectorizer() << 'y',
OneHotVectorizer() << ['workclass', 'education'],
TypeConverter(result_type='R4') << {'yi': 'y'},
Drop() << 'y',
FastLinearBinaryClassifier(max_iterations=1) << 'yi'
])
exp.fit(X, y, verbose=0)
prediction = exp.predict(X)
assert isinstance(prediction, pandas.DataFrame)
assert list(prediction.columns) == ['Score']
assert prediction.shape == (5, 1)
assert prediction.min() > 0.01
assert prediction.max() < 0.05
def test_syntax12_mixed2(self):
X = pandas.DataFrame(dict(education=['A', 'B', 'A', 'B', 'A'],
workclass=['X', 'X', 'Y', 'Y', 'Y'],
weight=[10., 1., 1., 1., 1.],
y=[1.1, 2.2, 1.24, 3.4, 3.4]))
exp = Pipeline(
[
OneHotVectorizer(
columns=[
'workclass', 'education']),
Concat(
columns={
'Feature': ['workclass', 'education']}),
FastTreesRegressor(
num_trees=5, feature='Feature', weight='weight') << {
Role.Label: 'y'}])
exp.fit(X, verbose=0)
assert exp.nodes[-1].feature_column_ == 'Feature'
assert exp.nodes[-1].label_column_ == 'y'
assert exp.nodes[-1].weight_column_ == 'weight'
# y is required here as well as weight.
# It is replaced by fakes values.
# The test does not fail but the weight is not taken into account.
X['y'] = -5
X['weight'] = -5
prediction = exp.predict(X)
assert isinstance(prediction, pandas.DataFrame)
assert list(prediction.columns) == ['Score']
assert prediction.shape == (5, 1)
def test_pipeline_with_no_columns(self):
trainData = pd.DataFrame({
"Sentiment": [0, 1, 1, 0, 1, 1],
"SentimentText": [
"this is train ", "review ", "sentence ", "an apple",
"sentence 22", "another one one one"
]
})
ppl = Pipeline([
NGramFeaturizer(word_feature_extractor=n_gram()),
LightGbmClassifier(min_data_per_leaf=1, min_data_per_group=1)
])
assert ppl is not None
# Bug 147697
info = ppl.get_fit_info(trainData[["SentimentText"]],
trainData["Sentiment"])
assert len(info) == 2
assert len(info[0]) == 3
ppl.fit(trainData[["SentimentText"]], trainData["Sentiment"])
ppl = Pipeline([
NGramFeaturizer(word_feature_extractor=n_gram()),
LightGbmClassifier(min_data_per_leaf=1, min_data_per_group=1)
])
assert ppl is not None
ppl.fit(trainData[["SentimentText"]], np.array(trainData["Sentiment"]))
def test_metrics_evaluate_binary(self):
np.random.seed(0)
df = get_dataset("iris").as_df()
df.drop(['Species'], inplace=True, axis=1)
df.Label = [1 if x == 1 else 0 for x in df.Label]
X_train, X_test, y_train, y_test = \
train_test_split(df.loc[:, df.columns != 'Label'], df['Label'])
lr = LogisticRegressionBinaryClassifier()
e = Pipeline([lr])
e.fit(X_train, y_train, verbose=0)
metrics, _ = e.test(X_test, y_test)
# TODO: debug flucations, and increase decimal precision on checks
assert_almost_equal(metrics['AUC'][0],
0.980,
decimal=1,
err_msg="AUC should be %s" % 0.980)
assert_almost_equal(metrics['Accuracy'][0],
0.632,
decimal=1,
err_msg="Accuracy should be %s" % 0.632)
assert_almost_equal(metrics['Positive precision'][0],
1,
decimal=1,
err_msg="Positive precision should be %s" % 1)
assert_almost_equal(metrics['Positive recall'][0],
0.125,
decimal=1,
err_msg="Positive recall should be %s" % 0.125)
assert_almost_equal(metrics['Negative precision'][0],
0.611,
decimal=1,
err_msg="Negative precision should be %s" % 0.611)
assert_almost_equal(metrics['Negative recall'][0],
1,
decimal=1,
err_msg="Negative recall should be %s" % 1)
assert_almost_equal(metrics['Log-loss'][0],
0.686,
decimal=1,
err_msg="Log-loss should be %s" % 0.686)
assert_almost_equal(metrics['Log-loss reduction'][0],
0.3005,
decimal=3,
err_msg="Log-loss reduction should be %s" % 0.3005)
assert_almost_equal(
metrics['Test-set entropy (prior Log-Loss/instance)'][0],
0.981,
decimal=1,
err_msg="Test-set entropy (prior Log-Loss/instance) should be %s" %
0.981)
assert_almost_equal(metrics['F1 Score'][0],
0.222,
decimal=1,
err_msg="F1 Score should be %s" % 0.222)
assert_almost_equal(metrics['AUPRC'][0],
0.966,
decimal=1,
err_msg="AUPRC should be %s" % 0.966)
def _test_schema_syntax_shift_df(self):
df = pandas.DataFrame(
data=dict(X1=[0.1, 0.2], X2=[0.1, 0.2], yl=[1, 0], tx=['e', 'r']))
exp = Pipeline(
[OneHotVectorizer() << 'tx',
FastLinearBinaryClassifier()])
exp.fit(df, 'yl')
def test_trees(self):
(train, label) = get_X_y(train_file, label_column, sep=',')
(test, label1) = get_X_y(test_file, label_column, sep=',')
pipeline = Pipeline([OneHotVectorizer() << categorical_columns,
FastTreesBinaryClassifier()])
pipeline.fit(train, label)
out_data = pipeline.predict(test)
check_accuracy(test_file, label_column, out_data, 0.65)
def test_default_label(self):
df = get_dataset("iris").as_df()
df.drop(['Species'], inplace=True, axis=1)
df.Label = [1 if x == 1 else 0 for x in df.Label]
# 1
pipeline = Pipeline([
ColumnConcatenator() << {
'Features': ["Petal_Length", "Sepal_Length"]
},
FastTreesBinaryClassifier(number_of_trees=2) << {
Role.Label: 'Label',
Role.Feature: 'Features'
}
])
model = pipeline.fit(df, verbose=0)
probabilities0 = model.predict_proba(df)
# 2
pipeline = Pipeline([
ColumnConcatenator() << {
'Features': ["Petal_Length", "Sepal_Length"]
},
FastTreesBinaryClassifier(number_of_trees=2) << {
Role.Feature: 'Features'
}
])
model = pipeline.fit(df, verbose=0)
probabilities = model.predict_proba(df)
assert_array_almost_equal(probabilities0, probabilities)
# 3
pipeline = Pipeline([
ColumnConcatenator() << {
'Features': ["Petal_Length", "Sepal_Length"]
},
FastTreesBinaryClassifier(number_of_trees=2)
])
model = pipeline.fit(df, verbose=0)
probabilities = model.predict_proba(df)
assert_array_almost_equal(probabilities0, probabilities)
# 4
pipeline = Pipeline([
ColumnConcatenator() << {
'Features': ["Petal_Length", "Sepal_Length"]
},
FastTreesBinaryClassifier(number_of_trees=2) << {
Role.Label: 'Label'
}
])
model = pipeline.fit(df, verbose=0)
probabilities = model.predict_proba(df)
assert_array_almost_equal(probabilities0, probabilities)
def test_trees_file(self):
pipeline = Pipeline([OneHotVectorizer() << categorical_columns,
FastTreesBinaryClassifier() << {
'Label': label_column}])
train_stream = FileDataStream(train_file, schema=file_schema)
pipeline.fit(train_stream, label_column)
test_stream = FileDataStream(test_file, schema=file_schema)
out_data = pipeline.predict(test_stream)
check_accuracy(test_file, label_column, out_data, 0.65)
