Understanding time series components

There are three components of time series that are key to understanding time-related data. They are trend, seasonality, and noise. Let's look at each of them in the context of our EU unemployment data.

Trend

The trend can be defined as the long-term tendency of the time series data—the fact that, on average, the values tend to increase or decrease over a period of time. Looking at our plot, we can identify three distinct trends:

A downward trend from 2005 until 2008 (less people unemployed on a year-on-year basis); an upward trend starting in 2008 and manifesting until 2013 (unemployment rose on average); and again, a downward trend between 2013, all the way until the end of 2017 (the number of people without work constantly decreased).

Seasonality

Seasonality is a regularly repeating pattern of highs and lows that is related to calendar time; that is, it's directly influenced by seasons, quarters, months, and so on. Think, for instance, about the electricity...