Hands-On Ensemble Learning with Python

Establishing a baseline

In order to establish a baseline, we will try to model the data using linear regression. Although it is a time series, we will not directly take time into account. Instead, we will utilize sliding windows of size S to generate features at each time point and use those features to predict the next point. Next, we will move the window one step forward in time to include the true value of the data point we predicted and discard the oldest data point inside the window. We will continue this process until all data points have been predicted. This is called walk-forward validation. One drawback is that we cannot predict the first S data points, as we do not have enough data to generate features for them. Another point of ...