def MyReport(model, model_Name, list_data, tune=True):
data_x, data_y, x_train, x_test, y_train, y_test = list_data
# Training
model.fit(x_train, y_train)
modelCV = model
if tune:
modelCV = model.best_estimator_
# General Report
y_predCV = modelCV.predict(x_test)
print(classification_report(y_test, y_predCV))
# Plot Confusion Matrix
le_result = loadLabelEncoder('LE/result.npy')
class_names = le_result.classes_
cnf_matrix = confusion_matrix(y_test, y_predCV)
plot_confusion_matrix(cnf_matrix,
classes=class_names,
title=model_Name +
' Confusion matrix, without normalization')
# ROC curve
try:
y_score = modelCV.decision_function(x_test)
plot_ROC_curve(y_test,
y_score,
title=model_Name + ' ROC curve',
class_names=class_names)
except:
print(
"ROC curve is not available because model does not have decision_function method"
)
# 10-fold-test error
scores = cross_val_score(modelCV, data_x, data_y, cv=10)
print("10-fold cross validation mean square error: ", np.mean(scores))
return modelCV
data = data_.iloc[:, 2:].copy()
# Label Encoder (only do 1 time)
#le = preprocessing.LabelEncoder()
#le_tour = saveLabelEncoder(data['tournament'],'LE/tournament.npy')
#data['tournament'] = le_tour.transform(data['tournament'])
#le_result = saveLabelEncoder(data['result'],'LE/result.npy')
#data['result'] = le_result.transform(data['result'])
#x = np.concatenate((data['team_1'],data['team_2']), axis=0)
#le_country = saveLabelEncoder(x,'LE/country.npy')
#data['team_1'] = le_country.transform(data['team_1'])
#data['team_2'] = le_country.transform(data['team_2'])
#data['home_team'] = le_country.transform(data['home_team'])
# Load Label Encoder
le_result = loadLabelEncoder('LE/result.npy')
data['result'] = le_result.transform(data_['result'])
le_tour = loadLabelEncoder('LE/tournament.npy')
data['tournament'] = le_tour.transform(data['tournament'])
le_country = loadLabelEncoder('LE/country.npy')
data['team_1'] = le_country.transform(data['team_1'])
data['team_2'] = le_country.transform(data['team_2'])
data['home_team'] = le_country.transform(data['home_team'])
# Standard Scale
#data_test = data.iloc[:,4:].copy()
#data_test[data_test.columns[:-1]] = MinMaxScaler().fit_transform(data_test[data_test.columns[:-1]])
#data = data_test.copy()