A Hybrid Model of Holt-Wintor and Neural Network Methods for Automobile Sales Forecasting

摘要

Forecasting is a common statistical venture in commercial enterprise, in which it facilitates to inform decisions about the scheduling of manufacturing, transportation and provides a guide to long-term strategic planning. The automobile sales forecast plays a vital role in business strategy for generating profit for an automobile enterprise corporation. However, it is a very challenging process due to the high level of complexity and uncertainty involved within the competitive world. This study proposed a hybrid model the usage of an Adaptive Multiplicative Triple Exponential Smoothing Holt-Winters (AHW) method and Backpropagation Neural Networks (BPNNs) to forecast automobile sales. The Indian automobile sales statistics has been used for both training and testing purposes. The result of the proposed method outperforms than the single forecasting model in terms of automobile sales forecasting.