How to train a random forest with scikit-learn

Random forests combine predictions from many randomized decision trees, making them a useful baseline when tabular classification depends on nonlinear relationships or feature interactions. A fitted forest can predict without feature scaling, while its tree count and depth still need deliberate limits for memory use and model complexity.

The built-in breast cancer dataset provides numeric features and two class labels for a compact training run. A stratified split reserves 20 percent of the rows for testing, and fixed random states make the split and forest repeatable while the program is being checked.

Held-out predictions provide the evidence that matters after fitting. Reporting the fitted tree count with test-set accuracy, precision, recall, and F1 score for both classes prevents one aggregate score from hiding a weak class result.

1. Create train_random_forest.py with the imports and stratified dataset split.

   train_random_forest.py

   from sklearn.datasets import load_breast_cancer
   from sklearn.ensemble import RandomForestClassifier
   from sklearn.metrics import classification_report
   from sklearn.model_selection import train_test_split
    
    
   data = load_breast_cancer()
   X_train, X_test, y_train, y_test = train_test_split(
       data.data,
       data.target,
       test_size=0.2,
       stratify=data.target,
       random_state=42,
   )

   stratify=data.target keeps both diagnosis classes represented in the training and test subsets. The integer random_state selects the same rows again when the inputs are unchanged.

2. Append the random forest definition after the split.

   forest = RandomForestClassifier(
       n_estimators=200,
       max_depth=5,
       random_state=42,
       n_jobs=-1,
   )

   n_estimators=200 sets the forest size, while max_depth=5 limits how deep each tree can grow. n_jobs=-1 lets fitting use all available CPU cores.

3. Append the fitting call after the forest definition.

   forest.fit(X_train, y_train)

4. Append the held-out prediction call after fitting.

   predictions = forest.predict(X_test)

5. Append the model and classification reporting after the prediction call.

   print(f"training rows: {X_train.shape[0]}")
   print(f"test rows: {X_test.shape[0]}")
   print(f"fitted trees: {len(forest.estimators_)}")
   print(f"held-out accuracy: {forest.score(X_test, y_test):.3f}")
   print("classification report:")
   print(
       classification_report(
           y_test,
           predictions,
           target_names=data.target_names,
       )
   )

6. Run the completed random forest training script.

   $ python3 train_random_forest.py
   training rows: 455
   test rows: 114
   fitted trees: 200
   held-out accuracy: 0.956
   classification report:
                 precision    recall  f1-score   support
   
      malignant       0.95      0.93      0.94        42
         benign       0.96      0.97      0.97        72
   
       accuracy                           0.96       114
      macro avg       0.96      0.95      0.95       114
   weighted avg       0.96      0.96      0.96       114

   fitted trees: 200 confirms that training created the requested forest size. The classification report comes from predictions for 114 held-out rows and exposes recall for both malignant and benign cases.