Summary

In this chapter, we have covered the classical (statistical) approaches to time series analysis and forecasting. We learned how to decompose any time series into trend, seasonal, and remainder components. This step can be very helpful in getting a better understanding of the explored time series. But we can also use it directly for modeling purposes.

Then, we explained how to test if a time series is stationary, as some of the statistical models (for example, ARIMA) require stationarity. We also explained which steps we can take to transform a non-stationary time series into a stationary one.

Lastly, we explored two of the most popular statistical approaches to time series forecasting—exponential smoothing methods and ARIMA models. We have also touched upon more modern approaches to estimating such models, which involve automatic tuning and hyperparameter selection.

In the next chapter, we will explore ML-based approaches to time series forecasting.

...