Machine learning model evaluation key concepts (Part 1)

When I started my first data science project, I just dived into a dataset which had a number of features and the labels. I came across a few tutorials on how to write the code in python and create models using the machine learning algorithms for classification tasks. I just briefly went through how the algorithms worked and without thinking much, created the models, ran the test set with the models and evaluated just by looking at the accuracy. Then I was like what’s next? How do I know how to finalize my model and how do I know if this is a good model? I kept hearing terms like confusion matrix, precision, recall and F1-score. I kept getting discouraged when I thought that I have to learn so many technical terms which were so “alien” to me.

I am into the 2nd year of my data science journey and I can say that I am now better at understanding all those terms that seemed intimidating to me earlier. In this article, I would like to share my understanding of those “alien” terms so that it will help someone who feels what I felt when I just started.

Machine learning (ML) is a subset of Artificial Intelligence (AI) that allows systems the ability to automatically learn and improve from an experience without being explicitly programmed. So, when we are done building an ML model, the question that pop is if our model is good enough. How do we evaluate that? There are a few metrics we can consider. If we understand these matrices well, we can pretty much understand a lot about the model’s…