Evaluating Machine Learning Classification Models and Sampling for Classification

Once we have some classification models trained to predict our target variable, we need a way to compare them and choose the best one. One way to compare models is to use metrics such as accuracy and others. In classification, we can often find that our classes or targets are imbalanced. We can improve the performance of ML classification algorithms by means of sampling techniques, such as oversampling and undersampling. In this chapter, we will learn about ways to evaluate our classification models and sampling methods:

• How to evaluate the performance of our algorithms (performance metrics)
• Sampling imbalanced data for classification

Let's start with metrics for comparing ML classification algorithms.