Optimizing a forecasting model with hyperparameter tuning

You trained different regression models using default parameter values in the previous recipe. A common term for such parameters is hyperparameters, as these are not learned by the model but instead supplied by the user, influencing the model's architecture and behavior.

In this recipe, you will examine how you can find optimal hyperparameter values for the KNN Regresssor (from the previous recipe). You will perform a cross-validated grid search using sktime's ForecastingGridSearchCV.

You have performed a grid search in the Forecasting univariate time series data with non-seasonal ARIMA recipe from Chapter 10, Building Univariate Time Series Models Using Statistical Methods. Similarly, you were introduced to different automated methods for finding optimal hyperparameters in auto_arima under the Forecasting time series data using auto_arima recipe in Chapter 11, Additional Statistical Modeling Techniques for...