def test_lm_cv_grid():
cv = ml.get_default_cv("linear_model", "cl", search_type="grid")
cv.param_grid = {
"ml__penalty": ["l1"],
"ml__C": [1e-5, 1e-3, 1e-1]
}
_test_basic_flow_holdout_pandas1(cv, True)
_test_basic_flow_holdout_pandas2(cv, True)
_test_basic_flow_cv_pandas(cv, True)
def test_xgboost_cv_grid():
cv = ml.get_default_cv("xgboost", "rg", search_type="grid")
cv.param_grid = dict(
ml__colsample_bynode=[0.1],
ml__learning_rate=[0.01],
ml__ratio_min_child_weight=[None, 0.005, 0.01]
)
_test_basic_flow_holdout_pandas1(cv, False)
_test_basic_flow_holdout_pandas2(cv, False)
_test_basic_flow_cv_pandas(cv, False)
def test_random_holdout(setup):
df_training, df_validation, df_test, feature_columns = setup
is_cl = False
target_column = utils.get_target_column(is_cl)
cv_obj = ml.get_default_cv("linear_model", "rg", "r2")
model = cv_obj.fit_holdout_pandas(df_training,
target_column,
feature_columns,
df_validation=df_validation)
analyzer = ml.CVAnalyzer(model.estimator)
_basic_flow(analyzer)
def test_random_forest_cv_grid():
cv = ml.get_default_cv("random_forest", "cl", search_type="grid")
cv.param_grid = dict(
ml__max_depth=[2],
ml__n_estimators=[100],
ml__max_features=["auto"],
ml__min_samples_leaf=[0.01, 0.05, 0.1]
)
_test_basic_flow_holdout_pandas1(cv, True)
_test_basic_flow_holdout_pandas2(cv, True)
_test_basic_flow_cv_pandas(cv, True)
def test_lightgbm_cv_grid():
cv = ml.get_default_cv("lightgbm", "cl", search_type="grid")
cv.param_grid = {
"ml__num_leaves": [10],
"ml__colsample_bytree": [0.1],
"ml__learning_rate": [0.01],
"ml__min_child_samples": [0, 20, 100]
}
_test_basic_flow_holdout_pandas1(cv, True)
_test_basic_flow_holdout_pandas2(cv, True)
_test_basic_flow_cv_pandas(cv, True)
def test_grid_holdout(setup):
df_training, df_validation, df_test, feature_columns = setup
is_cl = True
target_column = utils.get_target_column(is_cl)
cv_obj = ml.get_default_cv("linear_model", "cl", search_type="grid")
cv_obj.parameter_grid = {
"ml__penalty": ["l1", "l2"],
"ml__C": [1e-5, 1e-3, 1e-1]
}
model = cv_obj.fit_holdout_pandas(df_training,
target_column,
feature_columns,
ratio_training=0.8)
analyzer = ml.CVAnalyzer(model.estimator)
_basic_flow(analyzer)
def test_random_forest_cv_random():
cv = ml.get_default_cv("random_forest", "rg", search_type="random")
cv.n_iter = 5
_test_basic_flow_holdout_pandas1(cv, False)
_test_basic_flow_holdout_pandas2(cv, False)
_test_basic_flow_cv_pandas(cv, False)
def test_xgboost_cv_random():
cv = ml.get_default_cv("xgboost", "cl", search_type="random")
cv.n_iter = 5
_test_basic_flow_holdout_pandas1(cv, True)
_test_basic_flow_holdout_pandas2(cv, True)
_test_basic_flow_cv_pandas(cv, True)
def test_lm_cv_random():
cv = ml.get_default_cv("linear_model", "cl", search_type="random")
cv.n_iter = 2
_test_basic_flow_holdout_pandas1(cv, True)
_test_basic_flow_holdout_pandas2(cv, True)
_test_basic_flow_cv_pandas(cv, True)