"max_depth": [2, 3, 5, 7, 10],
"max_features":[2, 3],
"n_estimators":[50 ,100, 150, 175, 200, 250, 300]}
xgb_grid_search = GridSearchCV(estimator=XGBClassifier(),
param_grid=param_grid,
cv=5, verbose=2, n_jobs=-1)
xgb_grid_search.fit(X_train, y_train)
xgb_grid_search.best_params_
XGB_tuned = XGBClassifier(n_estimators=200,
learning_rate=0.25,
max_depth=20,
max_features=2,
random_state=42)
XGB_tuned.fit(X_train, y_train)
XGB_tuned.pred = XGB_tuned.predict(X_test)
accuracy = accuracy_score(y_test, XGB_tuned.pred)
fpr, tpr, thresholds = roc_curve(y_test, XGB_tuned.pred)
auc_score = auc(fpr, tpr)
print("Model : XGBoost (Grid Search Tuning)")
print("Accuracy :", accuracy)
print("AUC score:", auc_score)
fig, ax = plot_confusion_matrix(conf_mat=confusion_matrix(y_test, XGB_tuned.pred),
figsize=(8, 8),
show_absolute=True,
show_normed=False,
colorbar=False, class_names=["Open", "Close"])
plt.title("Confusion Matrix of XGB")
plt.show()
