How Forecasting Seasonality Improves Year-Round Predictability

Line Art

The climate, the holidays, the start of the school year, tax season: these are the events that drive our seasonal buying patterns, especially for seasonal needs and desires for things like gingerbread cookies,  vacation cottages, or pumpkin spice brew. If you live in the U.S., you may well have enjoyed a delicious turkey this month, with an eye on another for the Christmas holiday. Or will it be a ham, a goose, or a restaurant outing? This is where we give a thought to those doing demand forecasting and inventory planning in the U.S. and around the globe. Demand planners, whether grizzled industry insiders or grads fresh out of school, have a tough job right now when it comes to forecasting seasonality:

  • They are dealing with a mishmash of consumer shopping trends, unknown weather possibilities, and varying changing regional consumption patterns.
  • The holiday shopping season is in full swing, extending from Black Friday through a massive holiday season that lasts well into the new year.
  • Consumers shop through multiple channels — phones and laptops to brick-and-mortar stores — and have varying fulfillment demands.

Forecasting Seasonality: The Science of Predictability

We’ve written about how data, trends, market research, risk analysis, and other factors help planners to develop accurate forecasts. Seasonality is one of the most frequently tracked patterns used to improve the accuracy of demand forecasts across a wide range of scenarios. Analyzing seasonal patterns is useful to planners throughout the year, helping them to make the right calls on seasonal inventories through informed anticipation of what customers will buy, when, and how. Helpful methods for identifying and forecasting seasonality include:

  • Regression analysis: Helps determine the relationship between demand and demand drivers. It is especially useful in enabling planners to visualize trends and seasonality. Regression analysis includes linear regression, hyperbolic trend, logarithmic trend, square root trend, quadratic trend, power trend, exponential trend, and seasonal trends.
  • Spectral-analysis forecasting models: Filter “noise” out of cyclical data; sometimes they do a better job than simple curve fitting or moving averages. This technique enables planners to, for example, pull seasonal patterns out of data that also has a strong monthly pattern, showing seasonality that would be otherwise difficult to see. Spectral analysis forecasting methods include lowpass filter, spectral noise filter, and formant frequency. The modeling technique also enables planners to extend trends a year into the future.

In addition, planners can gain insights into quasi-seasonal events, such as those tied to other events, or select time periods that fall on different dates. Likewise, with adjustment tools for forecasting, planners can model certain seasonal products that have multiple iterations – another way to use the demand history of old, or base, products to forecast new products.

Forecasting seasonality is key to supply chain planning, inventory optimization, and supply chain optimization.

New to business forecasting? Get up to speed by reading our post Business Forecasting 101.