Trend, Seasonality & Cycles

A time series looks like one squiggly line — but it's really several patterns stacked on top of each other. Tease them apart and forecasting stops feeling like magic:

• Trend: the long-run drift — electricity demand creeping upward for decades as population grows
• Seasonality: repeats on a fixed calendar beat — demand peaks every evening, every summer
• Cycles: long swings with no fixed period — recessions, construction booms, fashion revivals
• Remainder: everything left once the first three are removed — noise, one-off shocks, pure randomness

Decomposition splits the line into these parts so you can model each separately and add the answers back.