MEASURING THE DEMAND FORECAST ACCURACY FOR SPARE PARTS Case: Wärtsilä Global Logistics Services
Yli-Kullas, Ulla (2016)
Yli-Kullas, Ulla
2016
Kuvaus
Opinnäytetyö kokotekstinä PDF-muodossa.
Tiivistelmä
Forecasting demand for spare parts is challenging due to large number of individual spare parts and the different demand patterns associated with them. The management of the spare parts is balancing between the expensive inventories, due to the high prices of the spare parts, and the customer satisfaction. Forecasting spare parts’ demand will allow lower inventories without dissatisfied customers.
The forecasts are never 100% accurate but are still the best estimates of future events. An important part of employing a demand forecast is to know the accuracy of it. The forecast accuracy can be measured in different ways and using different forecast error measures for different demand patterns is recommended.
The purpose of this thesis is to provide the case company with knowledge on the demand forecasting for spare parts as well as the error measures useful for the spare parts. The empirical part of the thesis consists of measure the accuracy of the forecast methods currently used in the case company. The SKUs with forecast were divided into SKUs with constant and intermittent demand. A representative sample of 5% was taken from both
groups in the calculations. Different error measures are used for different demand patterns as a single measure is not sufficient to all demand patterns.
The forecasts are never 100% accurate but are still the best estimates of future events. An important part of employing a demand forecast is to know the accuracy of it. The forecast accuracy can be measured in different ways and using different forecast error measures for different demand patterns is recommended.
The purpose of this thesis is to provide the case company with knowledge on the demand forecasting for spare parts as well as the error measures useful for the spare parts. The empirical part of the thesis consists of measure the accuracy of the forecast methods currently used in the case company. The SKUs with forecast were divided into SKUs with constant and intermittent demand. A representative sample of 5% was taken from both
groups in the calculations. Different error measures are used for different demand patterns as a single measure is not sufficient to all demand patterns.