How to compute precision, recall, accuracy and f1-score for the multiclass case with scikit learn?

Introduction

In multiclass classification, a model predicts one of multiple possible classes. Evaluating the performance of such models requires metrics beyond simple accuracy. This article will guide you through computing precision, recall, accuracy, and F1-score for multiclass classification using scikit-learn.

Understanding the Metrics

Precision

Precision measures the proportion of correctly predicted positive instances among all instances predicted as positive.

Recall

Recall measures the proportion of correctly predicted positive instances among all actual positive instances.

Accuracy

Accuracy is the overall proportion of correctly classified instances.

F1-Score

The F1-score is the harmonic mean of precision and recall, providing a balanced metric for evaluating performance.

Multiclass Evaluation with scikit-learn

Using `classification_report`

The `classification_report` function in scikit-learn provides a comprehensive report of all the metrics for each class.

from sklearn.metrics import classification_report

# Example usage
y_true = [0, 1, 2, 0, 1, 2]
y_pred = [0, 1, 2, 0, 2, 1]

print(classification_report(y_true, y_pred))

This code snippet will print a report including precision, recall, F1-score, and support for each class.

Using `precision_score`, `recall_score`, `accuracy_score`, and `f1_score`

You can also compute each metric individually using dedicated functions.

• precision_score(y_true, y_pred, average='macro'): Macro-averaged precision
• recall_score(y_true, y_pred, average='macro'): Macro-averaged recall
• accuracy_score(y_true, y_pred): Overall accuracy
• f1_score(y_true, y_pred, average='macro'): Macro-averaged F1-score

The `average=’macro’` argument calculates the metric as the unweighted mean of per-class scores, giving equal importance to all classes.

Table Summary

Conclusion

This article demonstrated how to compute precision, recall, accuracy, and F1-score for multiclass classification using scikit-learn. By understanding these metrics, you can effectively evaluate the performance of your multiclass classification models and make informed decisions about model selection and optimization.