Forecasting with Multiple Seasonality
This addresses forecasting challenges in applications with complex seasonal patterns, but it is incremental as it builds on existing ARMA models.
The paper tackles forecasting time series with multiple seasonality by proposing a two-stage method that generalizes seasonal ARMA models and uses lag order selection, showing excellent predictive performance compared to Facebook Prophet.
An emerging number of modern applications involve forecasting time series data that exhibit both short-time dynamics and long-time seasonality. Specifically, time series with multiple seasonality is a difficult task with comparatively fewer discussions. In this paper, we propose a two-stage method for time series with multiple seasonality, which does not require pre-determined seasonality periods. In the first stage, we generalize the classical seasonal autoregressive moving average (ARMA) model in multiple seasonality regime. In the second stage, we utilize an appropriate criterion for lag order selection. Simulation and empirical studies show the excellent predictive performance of our method, especially compared to a recently popular `Facebook Prophet' model for time series.