Core Machine Learning Metrics
All in one comprehensive guide through core machine learning model metrics. A few lessons learned from working with machine learning models.
Correctly evaluating model performance is a crucial task while working with machine learning. There are quite a few metrics that we may use to do so. For someone who just started the journey in this field that can be problematic — at least it was for me.
I will start with describing concepts like true/false negatives/positives as they are the base for more complex metrics. Then I will mention and explain metrics like accuracy, precision, recall, or calibration error. I will also explain the basics behind the confusion matrix along a short code snippet on how to build one.
Why?
Finding resources online and reading them is simple. Everyone can do it and I did it as well — but what I missed was an all in one glossary for all the stuff. This is my main motivation behind writing this text. I will describe all the metrics I came into contact with while working on my previous project.
I think that such a metrics glossary will be useful for all the people new to working with machine learning models.
