Developing a forecast modelling tool for a newly restructured fluid conveyance provider company

Abstract

The KLM Group trades roughly 15 000 products in the fluid conveyance industry. Managing inventory levels and production schedules is currently supported by a manual forecasting process. The company wishes to incorporate an automated forecasting process which can be used to estimate future monthly sales. A lead time of three months on most of their products renders out the necessity of a daily forecast. There are many different types of quantitative and qualitative forecasting methods that can be used. Presented in this document is a quantitative approach to forecasting depicting the different methods. These methods differ in terms of the type of historical data patterns they accommodate, i.e. trend, seasonal, constant, etc. The chosen methods presented in this document are most widely used in the industry as they cover the most essential data patterns. A model has been developed that uses each of these methods to estimate a forecast for every item. The forecast generated by the method that yields the smallest Mean Absolute Percentage Error (MAPE) is then chosen as the forecast for that item. This is referred to as the Best Fit Approach.