Course contentsShow
← Data Science
Lesson 768 of 2,14520. Time Series ModelsFree lesson

Holt-Winters Multiplicative Model

How the multiplicative Holt-Winters model handles seasonality that scales with the level of the series.

Holt-Winters Multiplicative Model

What You'll Learn

You'll learn how the multiplicative Holt-Winters model handles seasonality that grows or shrinks proportionally with the level of the time series.

Core Idea

The Holt-Winters Multiplicative Model is designed for time series where seasonal fluctuations change in size as the overall level of the series changes. Instead of adding fixed seasonal values (like the additive model does), it multiplies the level by seasonal factors.

Think of retail sales: if a store's baseline revenue doubles, the holiday season boost doesn't just add the same dollar amount—it doubles too. A 20% December bump applies to whatever the current sales level is. That's multiplicative seasonality.

How It Works

The multiplicative model uses three equations similar to the additive version, but with one key difference—the seasonal component is multiplied rather than added:

  • Level equation: Updates the average level, divided by the seasonal factor
  • Trend equation: Tracks the slope (same as additive)
  • Seasonal equation: Updates seasonal factors as ratios (not differences)

When forecasting, you multiply the (level + trend) by the appropriate seasonal factor, rather than adding it.

When to Use It

Choose the multiplicative model when:

  • Seasonal swings get larger as the series level increases
  • You see "V-shaped" or expanding seasonal patterns in your plot
  • The data represents percentages, ratios, or growth rates

Key Takeaway: The Holt-Winters Multiplicative Model captures seasonality that scales proportionally with the level of the series—when your peaks and valleys grow bigger together with the trend, multiply don't add.