Predicting Turnover in the Retail Fashion Industry – Integrating Data Mining and Heuristic Expert Information about Time Series

Forecasts are of vital importance in a business that undergoes seasonal changes as well as fashion trends with limited supplies. The straightforward approach is time series analysis. It here incorporates past turnover figures as well as orders planned for the following months.

The key challenge lays in the consideration of heuristic expert information about the series, which may explain outliers such as weather phenomena or delays in the supply chain. Experts e.g. need to be enabled to incorporate the likelihood of theses rare events into the overall and individual forecast in order to facilitate its acceptance.