What are forecast periods?

This article explains forecast intervals, and forecast periods in the AI & Analytics Engine's Demand Forecasting Machine Learning Solution.

What are forecast periods in demand forecasting?

A forecast period is a time interval over which predictions are made in future. It is a core component to consider in demand forecasting. Forecast periods can vary depending on the business nature, the product type and the forecast goal. For example

• A supermarket may need to predict the daily demand over the next month so that the delivery of their products can be arranged on time.
• A clothing retailer can forecast the sales for each product in a particular region for an upcoming quarter.
• A car manufacturer generally predicts the sales for different models over the next few years before production.

Forecast periods in the AI & Analytics Engine

In a demand forecasting app of the AI & Analytics Engine, the forecast period is defined as part of business requirements. Here the following factors are used to determine forecast periods

• Forecast horizon size (e.g. 4 weeks or 3 months)

• Forecast granularity (e.g. daily or monthly)

• Starting time of the first forecast period

• Lead time – time in advance required for businesses to respond to forecasted demands.

Below is an example of a forecast period over 3 months, starting from 2024-08-01, with lead time of 2 months on the 1st day of the month. Each recurring forecast includes the total sales of each month in the forecast period.

Example of a forecast period between 2024-08-01 and 2024-10-31

Example of how to specify forecast periods in the Engine

For an existing demand forecasting app, forecast results are typically organized by forecast periods on the app summary page. Particularly, users can zoom in a forecast period and check the details for each forecast horizon step.

Example of forecast results for different forecast periods in the Engine

Example of forecast results for a particular forecast period in the Engine