Comparison of Forcasting Models between the Value of using Credit and Debit Cards for the Payment of Goods and Services through the Holt-Winter and the Box-Jenkins Methods

Abstract

Currently, using credit and debit cards to pay for goods and services tends to be highly active. This research aimed to 1) create a model using the Holt-Winter and Box-Jenkins methods, and 2) compare the forecasting models between the value of using credit and debit cards for the payment of goods and services through the Holt-Winter and the Box-Jenkins. The data used spans from January 2010 to November 2023, totaling 167 months, and was collected from the Bank of Thailand's website. The data was divided into two sets: a learning set from January 2010 to February 2021, consisting of 133 months, and a testing set from March 2021 to November 2023. The data was collected by using the Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE) to compare the performance of the forecasting models.

The results of the study showed that the Box-Jenkins model was more suitable than the Holt-Winter model for both credit and debit cards. Financial institutions could use the results of the forecasting model in proactive public relations plans to encourage customers of financial institutions to use credit and debit cards more during times when usage is decreasing. Additionally, they could plan for software, hardware, and security system enhancements to support the increased number of customers using credit and debit cards in the future.