By Arizal Ibnu Haris
Student of S1 Statistika
Universitas Muhammadiyah Semarang (UNIMUS)
In the sale of a product, it is necessary to have an accurate estimate that will be very useful for XYZ stores in determining the stock of products in the future. The wrong estimation will lead to wrong planning which will lead to an increase in the cost of expenses for a company.
First of all, I think forecasting is the right way to predict the future demand for a product or service. There are many types of forecasting that can be done, such as in the case of this XYZ store using Single Exponential Smoothing and Double Exponential Smoothing which are used in predicting sales of XYZ store products. Single Exponential Smoothing used in short-term forecasting is usually used in a period of one month.
The accuracy of the forecast will be measured using the Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE) functions as well as standard deviation. Root Mean Square Error (RMSE) is commonly used to evaluate model performance in terms of its compatibility and true value. RMSE does not have a minimum standard value to recognize model performance unlike MAPE. The lower the MAPE score indicates that the forecasting model has good performance. In my opinion, forecasting using the Single Exponential Smoothing and Double Exponential Smoothing methods is easy to do in predicting a product in the short term. In this forecasting uses XYZ store product sales data from October 2017 - March 2018 which contains 24 entries corresponding to the number of weeks in that period.
In addition, This forecasting process for testing the Single Exponential Smoothing method produces a value of 20% MAPE, 4960 MAD, and 40920658 MSD. Whereas for testing the Double Exponential Smoothing method yielded 24% MAPE, 5686 MAD, and 47060161 MSD values. Based on the results of testing the 2 methods, the Single Exponential Smoothing method has the lowest MAPE score of 20%, so the Single Exponential Smoothing method can be categorized as an adequate forecasting method and furthermore it will be more suitable to use Single Exponential Smoothing as a forecasting method.
To sum up, I believe that the Single Exponential Smoothing method is the right method or model in forecasting the XYZ store case which can be seen from the test results, namely the Single Exponential Smoothing method has the lowest error with a MAPE score of 20% while the Double Exponential Smoothing method has a MAPE value of 24%. Therefore the Single Exponential Smoothing method is an adequate forecasting method to be used in this case.