Özet:

In this study, hybrid ARIMAX-ANN and SARIMAX-ANN sales forecasting models are proposed for a white goods wholesaler. White goods industry which is one of the durable goods sub-sector includes washing machines, dishwashers, refrigerators and small home appliances. In making forecasts, 46-month sales data of a white goods wholesaler are used. Factors influencing sales such as exchange rate, holidays, consumer confidence index (CCI), producer price index (PPI) and residential sales of the region are used as explanatory variables. The study contributes to the current literature by some aspects. First, there is no attempts applying the ARIMAX-ANN and SARIMAX-ANN hybrid models to forecast sales data in white goods industry. Second, the hybrid models combine the advantages of times series and ANN models. ARIMAX models are insufficient to solve complex nonlinear problems. On the other hand, ANN is sufficient to explain nonlinear relationships. On conclusion, use of hybrid models can increase the accuracy of the models.

Anahtar Kelimeler: