import numpy as np import pickle from tensorflow import keras # from sklearn.metrics import accuracy_score, f1_score, classification_report, confusion_matrix from RF.grappaBroskiSkibiddiLewandoski import HobaTitoMambo from Metrics import classification_report2, confusion_matrix2, accuracy_score2 #Download Data (X_train, y_train), (X_test, y_test) = keras.datasets.mnist.load_data() X_train = X_train / 255.0 X_test = X_test / 255.0 from skimage.feature import hog def extract_hog_features(X): features = [] for img in X: img_2d = img.reshape(28, 28) hog_features = hog( img_2d, orientations=9, pixels_per_cell=(4, 4), cells_per_block=(2, 2), block_norm='L2-Hys', feature_vector=True ) features.append(hog_features) return np.array(features) X_train_HOG = extract_hog_features(X_train) X_test_HOG = extract_hog_features(X_test) print("Initializing HobaTitoMambo Random Forest...") rf = HobaTitoMambo( n_trees=10, max_depth=15, min_samples_split=20, min_sample_leafs=5, max_features="sqrt", criterion="gini", random_state=42 ) print("Training model (this may take a few minutes)...") rf.fit(X_train_HOG, y_train) print("Predicting on test set...") predictions = rf.predict(X_test_HOG) conf_matrix = confusion_matrix2(y_test, predictions) print("\n--- Evaluation Results ---") print("\nConfusion Matrix:") print(conf_matrix) print("\nDetailed Classification Report:") print(classification_report2(y_test, predictions)) model_filename = 'rf_hog_model.pkl' print(f"\nSaving model to {model_filename}...") with open(model_filename, 'wb') as file: pickle.dump(rf, file) print("Model saved successfully!")