def test_feature_tools_transformer(self):
df = dsutils.load_bank()
df.drop(['id'], axis=1, inplace=True)
y = df.pop('y')
X_train, X_test = train_test_split(df.head(100), test_size=0.2, random_state=42)
ftt = FeatureGenerationTransformer(task='classification', trans_primitives=['add_numeric', 'divide_numeric'])
ftt.fit(X_train)
x_t = ftt.transform(X_train)
assert x_t is not None
def test_pipeline(self):
df = dsutils.load_bank()
df.drop(['id'], axis=1, inplace=True)
cross_cat = CrossCategorical()
X_train, X_test = train_test_split(df.head(100), test_size=0.2, random_state=42)
ftt = FeatureGenerationTransformer(task='classification', trans_primitives=[cross_cat])
preprocessor = general_preprocessor()
pipe = Pipeline(steps=[('feature_gen', ftt), ('processor', preprocessor)])
X_t = pipe.fit_transform(X_train)
assert X_t.shape == (80, 62)
def test_in_dataframe_mapper(self):
df = dsutils.load_bank()
df.drop(['id'], axis=1, inplace=True)
cross_cat = CrossCategorical()
X_train, X_test = train_test_split(df.head(100), test_size=0.2, random_state=42)
ftt = FeatureGenerationTransformer(task='classification', trans_primitives=[cross_cat])
dfm = DataFrameMapper(features=[(X_train.columns.to_list(), ftt)],
input_df=True,
df_out=True)
X_t = dfm.fit_transform(X_train)
assert X_t.shape == (80, 62)
def test_model(self):
df = dsutils.load_bank()
df.drop(['id'], axis=1, inplace=True)
X_train, X_test = train_test_split(df.head(1000), test_size=0.2, random_state=42)
y_train = X_train.pop('y')
y_test = X_test.pop('y')
def f():
return X_train, X_test, y_train, y_test
self.train_bankdata(f)
def test_feature_generation_with_selection(self):
df = dsutils.load_bank().head(1000)
df.drop(['id'], axis=1, inplace=True)
y = df.pop('y')
cross_cat = CrossCategorical()
ftt = FeatureGenerationTransformer(task='classification',
trans_primitives=['add_numeric', 'divide_numeric', cross_cat],
feature_selection_args={'ratio_select_cols': 0.2})
with pytest.raises(AssertionError) as err:
ftt.fit(df)
assert err.value == '`y` must be provided for feature selection.'
ftt.fit(df, y)
x_t = ftt.transform(df)
assert x_t.shape[1] == 35
def test_no_categorical(self):
df = dsutils.load_bank()
df.drop(['id'], axis=1, inplace=True)
df = df[['age', 'duration', 'previous', 'y']]
X_train, X_test = train_test_split(df.head(1000), test_size=0.2, random_state=42)
y_train = X_train.pop('y')
y_test = X_test.pop('y')
def f():
return X_train, X_test, y_train, y_test
self.train_bankdata(f)
def test_no_continuous(self):
df = dsutils.load_bank()
df.drop(['id'], axis=1, inplace=True)
df = df[['job', 'education', 'loan', 'y']]
X_train, X_test = train_test_split(df.head(1000), test_size=0.2, random_state=42)
y_train = X_train.pop('y')
y_test = X_test.pop('y')
def f():
return X_train, X_test, y_train, y_test
self.train_bankdata(f)
def test_feature_selection(self):
df = dsutils.load_bank().head(1000)
df.drop(['id'], axis=1, inplace=True)
y = df.pop('y')
cross_cat = CrossCategorical()
ftt = FeatureGenerationTransformer(task='classification',
trans_primitives=['add_numeric', 'divide_numeric', cross_cat])
ftt.fit(df)
x_t = ftt.transform(df)
fst = FeatureSelectionTransformer('classification', ratio_select_cols=0.2, reserved_cols=ftt.original_cols)
fst.fit(x_t, y)
assert len(fst.scores_.items()) == 99
assert len(fst.columns_) == 35
x_t2 = fst.transform(x_t)
assert x_t2.shape[1] == 35
def test_blend(self):
df = dsutils.load_bank()
df.drop(['id'], axis=1, inplace=True)
X_train, X_test = train_test_split(df.head(1000), test_size=0.2, random_state=42)
y_train = X_train.pop('y')
y_test = X_test.pop('y')
def f():
return X_train, X_test, y_train, y_test
_, hyper_model = self.train_bankdata(f)
samples = [t.space_sample for t in hyper_model.get_top_trails(3)]
blend_models = hyper_model.blend_models(samples, X_train, y_train)
score = blend_models.evaluate(X_test, y_test, ['AUC'])
assert score
def test_bankdata_catboost(self):
space = search_space_general(lightgbm_fit_kwargs=lightgbm_fit_kwargs, )
space.assign_by_vectors([2, 2, 1, 1, 0.031, 1, 3, 1])
estimator = HyperGBMEstimator(
'binary', space, cache_dir=f'{test_output_dir}/hypergbm_cache')
df = dsutils.load_bank()
df.drop(['id'], axis=1, inplace=True)
X_train, X_test = train_test_split(df.head(10000),
test_size=0.2,
random_state=42)
y_train = X_train.pop('y')
y_test = X_test.pop('y')
estimator.fit(X_train, y_train)
scores = estimator.evaluate(X_test, y_test, metrics=['accuracy'])
assert scores
print(scores)
def test_save_load(self):
df = dsutils.load_bank()
df.drop(['id'], axis=1, inplace=True)
X_train, X_test = train_test_split(df.head(1000), test_size=0.2, random_state=42)
y_train = X_train.pop('y')
y_test = X_test.pop('y')
def f():
return X_train, X_test, y_train, y_test
est, hypergbm = self.train_bankdata(f)
filepath = test_output_dir + '/hypergbm_model.pkl'
est.save(filepath)
assert os.path.isfile(filepath) == True
model = hypergbm.load_estimator(filepath)
score = model.evaluate(X_test, y_test, ['AUC'])
assert score
def main():
rs = RandomSearcher(search_space_general, optimize_direction=OptimizeDirection.Maximize)
hk = HyperGBM(rs, task='classification', reward_metric='auc',
cache_dir=f'{test_output_dir}/hypergbm_cache',
callbacks=[SummaryCallback(), FileLoggingCallback(rs, output_dir=f'{test_output_dir}/hyn_logs')])
df = dsutils.load_bank()
df.drop(['id'], axis=1, inplace=True)
X_train, X_test = train_test_split(df, test_size=0.8, random_state=42)
y_train = X_train.pop('y')
y_test = X_test.pop('y')
hk.search(X_train, y_train, X_test, y_test, max_trails=500)
best_trial = hk.get_best_trail()
print(f'best_train:{best_trial}')
estimator = hk.final_train(best_trial.space_sample, X_train, y_train)
score = estimator.predict(X_test)
result = estimator.evaluate(X_test, y_test, metrics=['auc', 'accuracy'])
print(f'final result:{result}')
def test_feature_tools_categorical_cross(self):
df = dsutils.load_bank()
df.drop(['id'], axis=1, inplace=True)
cross_cat = CrossCategorical()
X_train, X_test = train_test_split(df.head(100), test_size=0.2, random_state=42)
ftt = FeatureGenerationTransformer(task='classification', trans_primitives=[cross_cat])
ftt.fit(X_train)
x_t = ftt.transform(X_train)
assert len(set(x_t.columns.to_list()) - set(
['job', 'marital', 'education', 'default', 'housing', 'loan', 'contact', 'month', 'poutcome', 'y', 'age',
'balance', 'day', 'duration', 'campaign', 'pdays', 'previous', 'contact__marital', 'job__poutcome',
'contact__default', 'housing__month', 'housing__marital', 'loan__y', 'housing__job', 'loan__poutcome',
'month__poutcome', 'default__month', 'default__education', 'education__loan', 'education__housing',
'housing__loan', 'housing__poutcome', 'contact__housing', 'contact__loan', 'marital__y', 'contact__job',
'education__poutcome', 'default__marital', 'job__month', 'job__y', 'default__loan', 'education__marital',
'default__poutcome', 'default__y', 'contact__month', 'education__month', 'contact__education',
'contact__poutcome', 'job__marital', 'education__job', 'job__loan', 'contact__y', 'month__y',
'default__housing', 'default__job', 'poutcome__y', 'loan__marital', 'education__y', 'loan__month',
'marital__month', 'housing__y', 'marital__poutcome'])) == 0
def train_bankdata(self, data_partition):
rs = RandomSearcher(search_space_general, optimize_direction=OptimizeDirection.Maximize)
hk = HyperGBM(rs, task='classification', reward_metric='accuracy',
cache_dir=f'{test_output_dir}/hypergbm_cache',
callbacks=[SummaryCallback(), FileLoggingCallback(rs, output_dir=f'{test_output_dir}/hyn_logs')])
df = dsutils.load_bank()
df.drop(['id'], axis=1, inplace=True)
X_train, X_test, y_train, y_test = data_partition()
hk.search(X_train, y_train, X_test, y_test, max_trails=3)
best_trial = hk.get_best_trail()
estimator = hk.final_train(best_trial.space_sample, X_train, y_train)
score = estimator.predict(X_test)
result = estimator.evaluate(X_test, y_test)
assert len(score) == 200
return estimator, hk