Machine Learning 101 P7: How to choose and evaluate Regression models

Introduction

We’ve now covered some of the key models in the Regression family (with links). These include linear regression models:

• Simple linear regression
• Multiple linear regression
• Polynomial linear regression

And also non-linear regression models:

• Support Vector regression
• Random Forest Regression
• Decision tree regression

Regardless of the shiny names, the ultimate goal of regression models is to predict a continuous output based on a relationship between input variables and the continuous output. Based on this goal, we can 100% be certain that regression models should be used in case of predicting a continuous value with a hint of probable existing relationship(s) among key/input factors. However, since there is a huge selection of models to choose from, you must question when and where to use what model. That’s exactly what we will find out after reading this blog post. Before jumping right in model selection part, let’s first go through all key metrics to evaluate a regression model, which later on give clue on the most ideal model.

Regression model performance metric

R-squared

R-squared (also known as the coefficient of determination) is a statistical measure that helps…