Regression with random forest

In the previous recipe, Forecasting based on multiple regression, we learned how to use multiple variables in order to predict the variable that we are interested in. Sometimes, we have a lot of variables and we are not sure which ones we should choose as predictors. Also, predictor variables can be related among themselves in different ways, which complicates the setup of the model and the interpretation of the results. In recent years, random forest algorithm has gained popularity among analysts and data scientists, as they provide a solution to these problems. The random forest algorithm is based on decision tree approach. This approach can be used to predict both discrete class membership (classification) and exact values of a continuous variable (regression). In this recipe, we will cover the latter. Regression-based on decision tree works by...