Multi-peak Lifecycle Forecasting Model Based on Artificial Bee Colony Algorithm

Abstract

There are many issues in tradition lifecycle models especially when we apply them to the products whose sales curve has a feature of multiple peaks. Two of them are the bad performances on fitting the sales curve and forecasting the demand. In order to solve the first problem, we proposed a method of fitting in batches which divided an entire category into some launching batches and fitted the sales curve of every launching batch separately, and then we got the sales curve of the category by adding the sales curve of every batch up. In order to solve the second problem, we introduced the Artificial Bee Colony Algorithm and built a self-update lifecycle forecasting model. Tested by a real sales data in garment industry and comparing with the tradition method and model, the method and the forecasting model can realize a higher accuracy both in fitting and forecast.