What is the reason for not increasing accurac much By using PSO for features selection with RF for prediction compared to other methods such as SVM?
Define the fitness function using RF def fitness_function(features): if not features: return 0.0 Create individual classifiers clf= RandomForestClassifier(n_estimators=52,max_depth=20, criterion=”gini”,random_state=777) selected_feature_indices = [list(X.columns).index(feature) for feature in features] selected_features = X_train[:, selected_feature_indices] Predictions clf.fit(selected_features, y_train) `your text` y_pred = clf1.predict(X_test[:, selected_feature_indices]) `your text` return accuracy_score(y_test, y_pred) Lists to store progress data iteration_values = [] accuracy_values = [] […]