Pushing random forest boundaries – case study

Imagine you work for a bike rental company and your goal is to predict the number of bike rentals per day depending upon the weather, the time of day, the time of year, and the growth of the company.

Earlier in this chapter, you implemented a random forest regressor with cross-validation to obtain an RMSE of 945 bikes. Your goal is to modify the random forest to obtain the lowest error score possible.

Preparing the dataset

Earlier in this chapter, you downloaded the dataset df_bikes and split it into X_bikes and y_bikes. Now that you are doing some serious testing, you decide to split X_bikes and y_bikes into training sets and test sets as follows:

from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(X_bikes, y_bikes, random_state=2)

n_estimators

Start by choosing a reasonable value for n_estimators. Recall that n_estimators can be increased to improve accuracy...