El modelo ARIMA elegido debe compararse con un modelo de Holt-Winters.

Generalized Additive Models (GAMs) extend linear models by allowing non-linear smooth functions of predictors while maintaining interpretability