How to test machine learning models
Quick answer
To test machine learning models, use a separate
test dataset to evaluate performance metrics like accuracy, precision, and recall. Employ techniques such as cross-validation and confusion matrices to ensure your model generalizes well and avoids overfitting.PREREQUISITES
Python 3.8+pip install scikit-learn>=1.2Basic knowledge of machine learning concepts
Setup
Install the necessary Python package scikit-learn for model evaluation and metrics.
pip install scikit-learn>=1.2Step by step
This example shows how to train a simple classifier, split data, and test the model using accuracy and a confusion matrix.
from sklearn.datasets import load_iris
from sklearn.model_selection import train_test_split
from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import accuracy_score, confusion_matrix
# Load dataset
iris = load_iris()
X, y = iris.data, iris.target
# Split data into train and test sets
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
# Train model
model = RandomForestClassifier(random_state=42)
model.fit(X_train, y_train)
# Predict on test set
predictions = model.predict(X_test)
# Evaluate
acc = accuracy_score(y_test, predictions)
cm = confusion_matrix(y_test, predictions)
print(f"Accuracy: {acc:.2f}")
print("Confusion Matrix:\n", cm) output
Accuracy: 1.00 Confusion Matrix: [[10 0 0] [ 0 8 0] [ 0 0 12]]
Common variations
You can use cross-validation to better estimate model performance by splitting data multiple times. For large datasets, consider streaming evaluation or using different models like GradientBoostingClassifier.
from sklearn.model_selection import cross_val_score
from sklearn.ensemble import GradientBoostingClassifier
model = GradientBoostingClassifier(random_state=42)
scores = cross_val_score(model, X, y, cv=5, scoring='accuracy')
print(f"Cross-validation accuracies: {scores}")
print(f"Mean accuracy: {scores.mean():.2f}") output
Cross-validation accuracies: [1. 0.97 1. 0.97 1. ] Mean accuracy: 0.99
Troubleshooting
If your model shows very high training accuracy but low test accuracy, it is likely overfitting; try regularization or more data. If metrics are unexpectedly low, verify data preprocessing and label correctness.
Key Takeaways
- Always evaluate machine learning models on a separate test set to measure true performance.
- Use metrics like accuracy, precision, recall, and confusion matrices to understand model behavior.
- Cross-validation provides a more robust estimate of model generalization than a single train-test split.