Two Primary Strategies for Multi-period Time Series Forecasting
We need the weather forecasts for a week to plan trips. A manufacturer needs forecasts for several months to plan and coordinate with others. The pursuit for accurate predictions for the future has never ceased, from the ancient sages peering into celestial patterns to the cutting-edge algorithms of today.
From a data science technical perspective, there is a big methodological difference between forecasting for one period and forecasting for multiple periods, and the latter is harder. Forecasting multiple periods demands a model to capture the intricate dynamics over time. The good news is that, in the forecasting literature, there are already innovative modeling solutions that multi-period forecasts.
Let’s assume we do not know the past literature and need to start from scratch. How do we invent the strategies to provide multi-period forecasts? Maybe we can start with the one-period forecasting models like ARIMA. How do we extend it to multi-period? One way is to use the same model recursively. We get the one-period prediction from the model. We include it as the input to forecast the second period. Then we include the prediction of the second period as the input to forecast the third period. We can iterate through all the periods by including the predictions of the prior periods. This is the recursive…
