An ANN Approach to Demand Forecasting in Retail Trade in Bangladesh

Abstract

For a small retail trading chain, demand forecasting is the main driver to effectiveness and efficiency. However, as large number of varied models and products are marketed through a super market, several attributes affect forecasting. Because of these affecting parameters, nonlinearity arises. As a result, traditional forecasting approaches can not provide good estimation of demand. A fuzzy neural network analysis can provide better solution in this case. This research first analyzed the trend and seasonality patterns of a selected product in a retail trading chain in Bangladesh. Then demand was forecasted using traditional Holt-Winter's model. The same was done again using artificial neural network (ANN) with fuzzy uncertainty. Finally, the errors, measured in terms of MAPE, were compared for finding the best fitting forecasting approach. The research found that the error levels in Holt-Winter's approach are higher than those obtained through fuzzy ANN approach. This is because of influence of several factors on demand function in retail trading system. It was also observed that as forecasting period becomes smaller, the ANN approach provides more accuracy in forecast. An efficient and accurate demand forecasting system can play a major role in minimizing inventory costs of a super market. It must be noted that inventory cost is the single most important element of total cost for a super market. As such minimization of inventory cost should be of prime importance. Thus, an optimum forecast is essential (2). Although the market share of super stores in Europe and the USA is really significant in retail business, it is not the same in a developing country, like Bangladesh. Traditionally, the retail business in Bangladesh is dominated by small independent grocery stores. However, the recent trend in the capital of Bangladesh, the mega city of Dhaka, with a population of 15 million, has changed. Super markets are rapidly gaining market share in retail business. Driven by changing life style, strong growth of middle class group, coupled with favorable demographic pattern, organized large stores concept is expanding very fast. Currently there are 35 large super markets only in the city of Dhaka. Based on recent growth of super markets in Dhaka, it is expected that the overall retail market share of the super markets will increase rapidly in the next few years. As such an efficient inventory management system, coupled with an accurate forecasting system has become essential for such chain markets.