Double Weighted Moving Average: Alternative Technique for Chemicals Supplier’s Sales Forecast

Autor: Ma. Magdalena Chain Palavicini, Roberto Del Rio Soto, Ma. Del Rocio Castillo E., M. Javier Cruz Gómez
Rok vydání: 2016
Předmět:
Zdroj: International Journal of Business Administration. 7
ISSN: 1923-4015
1923-4007
DOI: 10.5430/ijba.v7n4p58
Popis: The proposed technique which has been called Double Weighted Moving Average (DWMA) was compared in performance with five already analyzed, Castillo, M. del R. et al. (2015), quantitative techniques: simple moving average, weighted moving average, exponential smoothing, trend projection and lineal regression; in order to improve the accuracy of chemicals supplier’s sales forecasts used to plan operations in Mexican industry. These forecasts are complex because; some Mexican companies handle up to 500 different materials and have up to 1000 customers with changing consumption patterns.The present DWMA forecasting technique was added to the system (algorithms and software) developed in the first part of the work, Castillo, M. del R. et al. (2016). Errors obtained with the five originally proposed techniques were compared versus the error obtained with the new technique.The system was applied to two Mexican chemicals suppliers. The new DWMA proposed technique showed, for the company one, better performance than the other five techniques, because most of its products have sales behavior with various patterns combined with seasonality. In the case of the company two, DWMA technique turned out to be the third of the six techniques used with lower error, because the behavior of most of its products presents random behavior sales data. The symmetric medium absolute percentage error, SMAPE, measurement is used on the system to measure each technique error. On the other hand, DWMA technique was better than the forecasting technique used in each company, in 56% for the first company products, who uses simple moving average, and in 81.8% of the second company products, who uses trend projection.
Databáze: OpenAIRE