Performance Metrics for Classification in Machine Learning: Understanding Accuracy, Precision, Recall and F1 Score

Introduction

In Machine Learning, one essential step is evaluating the performance of a model. For classification models, the Confusion Matrix serves as a fundamental instrument for evaluating the performance. The Confusion Matrix provides a visualization of the results of a model. Based on the information from the Confusion Matrix, some essential metrics can be calculated that provide an indication of the model performance. The key performance metrics are Accuracy, Precision, Recall and F1 Score. In this tutorial, we explore how to calculate these performance metrics based on the Confusion Matrix and how to interpret them.

Confusion Matrix

If you are not familiar with the Confusion Matrix, make sure to check out the following post:

Let's consider again the Binary Classification problem from this post. Suppose we have developed a Machine Learning model to classify images as cats or as dogs. In this example, we have the following two classes:

Positive: Cat
Negative: Dog

After training the model, we apply it to a set of test images to evaluate its performance. We have a total number of 40 test images, of which 26 are cats and 14 are dogs. We have already visualized the results of the model with a Confusion Matrix.

We obtain the following values from the Confusion Matrix:

True Positive (TP) = 20
True Negative (TN) = 10
False Positive (FP) = 4
False Negative (FN) = 6

Performance Metrices

There are some key performance metrics that can be calculated based on the Confusion Matrix. The most essential metrics are Accuracy, Precision, Recall and F1 Score.