def xgboost(data, target, test_data, test_target=None, features=None):
n = 406
print ('n = ' + str(n))
model = XGBClassifier()
model.fit(data, target)
prediction = model.predict(test_data)
cv_score = cross_val_score(model, data, target, cv=10, scoring=scorer)
# print(prediction)
# print(test_target)
print ('F1 score = ' + str(cv_score.mean()))
print ('F1 score std = ' + str(cv_score.std()))
# Feature importances into a dataframe
if features is not None:
feature_importances = pd.DataFrame({'feature': features, 'importance': model.feature_importances_})
# print(feature_importances.head())
validation.cal_rmse(prediction, test_target)
def gradient_boosting(data, target, test_data, test_target=None):
model = GradientBoostingClassifier(n_estimators=100, learning_rate=1.0, max_depth=24, random_state=0)
model.fit(data, target)
prediction = model.predict(test_data)
cv_score = cross_val_score(model, data, target, cv=10, scoring=scorer)
# print(prediction)
# print(test_target)
print ('F1 score = ' + str(cv_score.mean()))
print ('F1 score std = ' + str(cv_score.std()))
return validation.cal_rmse(prediction, test_target)
def random_forest(data, target, test_data, test_target=None, features=None):
# for n in range(407, 410):
n = 408 # 408 seems best with max_features 59 in these values
print ('n = ' + str(n))
model = RandomForestClassifier(n_estimators=n, random_state=10, n_jobs=-1, max_features=59)
model.fit(data, target)
prediction_fit = model.predict(data)
prediction = model.predict(test_data)
cv_score = cross_val_score(model, data, target, cv=10, scoring=scorer)
# print(prediction)
# print(test_target)
print ('F1 score = ' + str(cv_score.mean()))
print ('F1 score std = ' + str(cv_score.std()))
# Feature importances into a dataframe
# if features is not None:
# feature_importances = pd.DataFrame({'feature': features, 'importance': model.feature_importances_})
# # print(feature_importances.head())
validation.cal_rmse(prediction, test_target)
return prediction_fit, prediction
def decision_tree_reg(data, target, test_data, test_target=None):
model = DecisionTreeRegressor()
model.fit(data, target)
prediction = model.predict(test_data)
cv_score = cross_val_score(model, data, target, cv=10, scoring=scorer)
# print(prediction)
# print(test_target)
print ('F1 score = ' + str(cv_score.mean()))
print ('F1 score std = ' + str(cv_score.std()))
return validation.cal_rmse(prediction, test_target)
def kneighbors_reg(data, target, test_data, test_target=None):
model = KNeighborsRegressor()
model.fit(data, target)
prediction = model.predict(test_data)
return validation.cal_rmse(prediction, test_target)
def kneighbors(data, target, test_data, test_target=None):
model = KNeighborsClassifier()
model.fit(data, target)
prediction = model.predict(test_data)
return validation.cal_rmse(prediction, test_target)
def logistic_regression(data, target, test_data, test_target=None):
model = LogisticRegression()
model.fit(data, target)
prediction = model.predict(test_data)
return validation.cal_rmse(prediction, test_target)
def linear_regression(data, target, test_data, test_target=None):
model = LinearRegression()
model.fit(data, target)
prediction = model.predict(test_data)
validation.cal_rmse(prediction, test_target)
return prediction