Prognostisering : En fallstudie för att jämföra Holt-Winters metod och Regressionsanalys

Detta är en Kandidat-uppsats från Mittuniversitetet/Avdelningen för informationssystem och -teknologi; Mittuniversitetet/Avdelningen för informationssystem och -teknologi

Sammanfattning: The aim of this study has been to compare the accuracy of prognosis tools and how well they do with regards to seasonal variation and known deviations in order to decide which method is more suitable for companies with the goal of optimizing their staffing. The study has used Systembolaget as a case to compare two methods of prognosis. Data has been retrieved from Systembolaget and includes the company's sales. The methods that have been compared are Holt-Winters method and regression analysis. This has been done by applying the methods in order to generate a forecast and then evaluating the result through MAPE. Application of these methods was done with Excel and SPSS. The study showed that, under the present circumstances, both methods on average gave equally accurate forecasts. Holt-Winter's method however, forecasted more accurately week by week, while the regression analysis projection on average yielded almost equal percentage error. The conclusion that was generated by the study was that Holt-Winter's method is preferable for companies with this type of sale if the purpose is to generate a weekly accurate forecast. On the other hand, the conclusion may vary if methods are optimized and more data is available.

  HÄR KAN DU HÄMTA UPPSATSEN I FULLTEXT. (följ länken till nästa sida)