An Improved Demand Forecasting with Limited Historical Sales Data

In retail business, the biggest challenge is to forecast the demand for new products with limited historical data. In this business, companies routinely introduce new products to increase their value in different aspects like market share, sales, revenue, margin, etc. The new product may be completely a new product or line extension product. This creates a major challenge for the forecaster. There are various statistical approaches which are used to forecast the demand. But those models work well if and only if data is sufficient and consistent. Otherwise, they produce inaccurate results. Very few approaches are available to forecast the demand for the Geo-Prods if historical data is limited. In this paper, developed a technique to identify similar Geo-Prods which are having sufficient sales history and to improve the forecast accuracy by using k-Nearest Neighborhood method and exponential weights. Results obtained from the proposed method have shown that the method is accurate and effective for forecasting the demand of short sales history Geo-Prods.