Added best_random_state_in_random_forest.py file an algorithm to find best random state of random forest classifier

machine_learning/best_random_state_in_random_forest.py

@@ -0,0 +1,58 @@
+import pandas as pd
+import warnings
+from sklearn.model_selection import train_test_split
+from sklearn.metrics import accuracy_score
+from sklearn.ensemble import RandomForestClassifier
+from sklearn.preprocessing import StandardScaler
+
+warnings.filterwarnings('ignore')
+
+
+def find_best_random_state(data: pd.DataFrame, target_column: str, iterations: int = 200) -> int:
+    """
+    Find the best random state for the Random Forest Classifier that maximizes accuracy.
+
+    Args:
+        data (pd.DataFrame): The dataset containing features and target variable.
+        target_column (str): The name of the target column in the dataset.
+        iterations (int): Number of random states to test. Default is 200.
+
+    Returns:
+        int: The random state that provides the best accuracy.
+    """
+    # Split dataset into predictors and target
+    predictors = data.drop(target_column, axis=1)
+    target = data[target_column]
+
+    # Split dataset into train and test sets
+    X_train, X_test, y_train, y_test = train_test_split(predictors, target, test_size=0.20, random_state=0)
+
+    # Scale features
+    scaler = StandardScaler()
+    X_train_scaled = scaler.fit_transform(X_train)
+    X_test_scaled = scaler.transform(X_test)
+
+    max_accuracy_rf = 0
+    best_random_state = 0
+
+    # Loop through specified random states
+    for random_state in range(iterations):
+        rf = RandomForestClassifier(random_state=random_state)
+        rf.fit(X_train_scaled, y_train)
+        y_pred_rf = rf.predict(X_test_scaled)
+
+        current_accuracy = round(accuracy_score(y_test, y_pred_rf) * 100, 2)
+        if current_accuracy > max_accuracy_rf:
+            max_accuracy_rf = current_accuracy
+            best_random_state = random_state
+
+    print(f"The best random state is: {best_random_state} with an accuracy score of: {max_accuracy_rf} %")
+    return best_random_state
+
+
+if __name__ == "__main__":
+    # Load dataset
+    dataset = pd.read_csv("heart.csv")
+
+    # Find the best random state
+    best_state = find_best_random_state(dataset, target_column="target")