Trend and seasonality decomposition

We have seen how there are natural trends and cycles that are shown from the time-series data. By visualizing charts and utilizing moving averages, we were able to identify overall trends and seasonalities. However, there are more statistical approaches to decomposing time-series data into trend and seasonality components. Largely, there are two main ways to do time-series decomposition:

• Additive: As the name suggests, the additive time-series decomposition method decomposes the data into trend, seasonality, and error (which is the component that cannot be explained by the overall trend and seasonality) so that when they are summed together, it can reconstruct the original time-series data:

Y_t = Trend_t + Seasonality_t + Error_t

• Multiplicative: On the other hand, the multiplicative time-series decomposition method decomposes the data into trend, seasonality, and error in a way that when they are multiplied together, the...