Optimising Double Exponential Smoothing for Sales Forecasting Using The Golden Section Method

Abstract

To achieve maximum profits and a satisfying impression on consumers, companies are required to have the right strategy in selling their products. In determining the right strategy, it requires the availability of accurate information that can be analyzed to determine a sales strategy so that it can increase the number of sales and generate large profits, namely by forecasting. In the Double Exponential Smoothing method, the problem that arises is determining the optimum α parameter value to provide the smallest size of forecasting error, which is sought using the trial and error method, so it requires quite a lot of time. To overcome this problem, a non-linear optimization algorithm using the Golden Section algorithm is used. The Golden Section algorithm is an algorithm that uses the principle of reducing the boundary area α which might produce a minimum objective function value. It is hoped that this forecasting design will be able to provide information that will help the company take decisions or steps in providing stock of goods for sale so that there will be no overstock in the warehouse and can increase Dewaayu Shop's profits.  based on the test results, the value of  MAPE value is obtained of 21.59579369% and RMSE value of 2.42465034.