Time Series Decomposition breaks down a time series into several components to better understand the underlying patterns:
- Trend Component: The long-term progression of the series (increasing, decreasing, or stable).
- Seasonal Component: Repeating patterns that occur at regular intervals (daily, weekly, monthly, quarterly, etc.).
- Residual Component: Random variations that cannot be attributed to trend or seasonality.
Interactive Features:
- Component Toggles: Turn on/off trend, seasonal, and residual components to see how each contributes to the original series.
- Decomposition Model: Switch between additive (y = trend + seasonality + residual) and multiplicative (y = trend × seasonality × residual) models.
- Seasonality Strength: Adjust how pronounced the seasonal pattern appears in the data.
- Trend Direction: Change the trend from decreasing to flat to increasing to see how it affects the decomposition.
- Hover Interactions: Hover over points in any component to see corresponding points in all components.
Experiment with the multiplicative model when dealing with data where seasonal variation increases with the trend level, like retail sales that grow each year but still have stronger holiday seasons.