Chapter 7: Forecasting the Temperature with ARIMA and SARIMA Models
In the previous chapter, we talked about our first forecasting use case, with fairly uncomplicated statistical techniques. In this chapter, we will continue to implore statistical techniques to generate forecasts, but we will move on to the very popular and robust ARIMA and SARIMA models. ARIMA, and its big brother SARIMA, are acronyms that stand for (Seasonal) Auto-Regressive Integrated Moving-Average. You can think of it in four parts:
- AR: Auto-regressive
- I: Integrated
- MA: Moving average
- S: Seasonal
Each one of these terms represents a separate technique that is combined with the (S)ARIMA model. In this chapter, you’ll learn about strong and weak stationarity, how to induce this in your data, the ARIMA and SARIMA models, and how to derive their hyperparameters from auto-correlation and partial auto-correlation plots.
In this chapter, we’ll cover the following topics:
...
