def tune_hyperparams_scikit(self, model, X_train, y_train):
"""
Tuning parameters of model using scikit learn GridSearch
"""
print("Started CV task at " +
datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S"))
# create gini scorer
measurer = Measurement()
gini_score = measurer.get_gini_scorer()
params = [{
"learning_rate": [0.01],
"n_estimators": [100, 200],
"seed": [100],
"max_depth": [2],
"min_child_weight": [1, 5],
"subsample": [1]
}]
print("Running GridSearch")
gscv = GridSearchCV(model,
params,
cv=2,
n_jobs=-1,
scoring=gini_score,
verbose=3,
error_score=0)
gscv.fit(X_train, y_train)
best_model = gscv.best_estimator_
# save best model
save_model(best_model, MODEL_DIR, prefix="CV5-bestModel-")
# save CV results
gscv_resutls = pd.DataFrame(gscv.cv_results_)
gscv_resutls.to_csv("GridSearchRestest.csv", index=False)
print("Finished CV task at " +
datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S"))
