Research on Retailer Order Forecast Based on Improved Exponential Smoothing Method

Aiming at the shortcomings of the traditional exponential smoothing method in the prediction of retailer orders, an improved three - order exponential smoothing method is proposed. The optimal smoothing coefficient a is searched by the hybrid leapfrog algorithm. With the increase of the data, the a value also changes dynamically, which improves the efficiency of prediction and improves the accuracy of prediction. According to the number of retail sales data, dynamically change the size of the initial value of the smooth, reducing the value of the forecast on the smooth initial value of the degree of dependence. The experimental results show that the non - linear and non - regular characteristics of the retailer order quantity are obvious. The improved three - order exponential smoothing method can be used to forecast the retailer's order and the prediction accuracy is very high. The improved exponential smoothing method has a strong ability to adapt to the non-linear and non-regular nature of retailer orders.