Time series are a very common type of data—they can be used to represent key business metrics such as financial prices, resource usage (energy, water, raw materials, and so on), weather patterns, or macroeconomic trends—and the list could go on and on. The particularity of time series is that the data has to be collected at regular intervals, and the key aspect of time series analysis is exploring ways that allow us to understand past values so that we can predict future ones.
One powerful approach is to decompose a time series into a combination of trend, cycle, seasonality, and irregular (also called error or noise). We learned how to do this in this chapter while we analysed the EU's unemployment data. We started by learning to compute the trend component by means of moving averages. Then, we applied multiplicative series decomposition formulas to calculate seasonality and error, and we also applied basic forecasting methods to predict future values. In the process, we learned...