How do We Quantify the Quality of Our Predictions? Part I

We have all worked on different kinds of Machine learning models, and each model needs to be evaluated in different ways. From the initial data that is provided to the outcome and the way, we as the users want to use it. A classification model would require a different metric for model evaluation as compared to a regression model or a Neural Net, and it's important to know and understand which metric to use and when. Here in this series, we go through some of these metrics, starting from the basic and the most commonly used ones to the application-specific and complex metrics that we can use. We will be starting with the basic metrics from sklearn and progress towards the more complicated metrics after that. Accuracy Score: In classification, the number of labels predicted for a sample versus the corresponding set of labels in y_true can be coined as accuracy.