Additive vs Multiplicative Models
What you'll learn: How to distinguish between additive and multiplicative seasonal patterns and choose the right decomposition model for your time series.
The Core Difference
In time series analysis, you must decide how the components (trend, seasonality, and irregular fluctuations) combine to form your observed data. There are two fundamental ways they interact:
Additive models assume components are added together:
Observed = Trend + Seasonality + Irregular
Multiplicative models assume components are multiplied:
Observed = Trend × Seasonality × Irregular
When to Use Each Model
Use Additive When:
- Seasonal fluctuations remain constant in absolute size regardless of the trend level
- Example: Daily temperature variations stay around ±10°F throughout the year, whether the baseline is 30°F or 70°F
Use Multiplicative When:
- Seasonal fluctuations grow or shrink proportionally with the trend
- Example: Retail sales with 20% holiday bumps—if baseline sales double, the holiday spike doubles too
The Visual Test
Plot your time series and ask: "Do the peaks and valleys get bigger as the trend increases?" If yes, use multiplicative. If the seasonal swings stay roughly the same height, use additive.
Think of it like this: additive models add the same "dollop" of seasonality regardless of scale (like adding a fixed amount of sugar to any coffee). Multiplicative models apply a percentage change (like a 15% discount—larger on expensive items).
Key Takeaway: Choose additive models when seasonal variations are constant in magnitude; choose multiplicative models when seasonal variations scale proportionally with the trend level.