Using case-based reasoning to improve the performance of Bass model

Forecasting model is a systematical tool to improve the performance of the demand forecasting in supply chain. With the incessant innovation of technology, new products have been developed continuously. Therefore, improving the forecasting accuracy of the new product has become a primary concern for many companies. Most of them in SCM care how to predict the new products sales in the future. Bass model was usually used to obtain the trend of the new product forecast by many reach in the past. Case-based Reasoning (CBR) is a method which can consider the past case to transform the useful historical experience into the case at present. To address this concern, we developed a demand forecasting methodology that combines CBR and common forecasting model. In the first stage, we use Bass model to obtain the new product sales forecasts. In the second stage, we obtain second forecast using CBR inference method to find the most similar curve in the past series. The final stage, we combine the forecast from Bass and CBR to development the hybrid forecast. Several cases are used to demonstrate the hybrid model how to improve the performance of new product forecast. The results show that the proposed hybrid model can effectively improve the accuracy of new product forecast.