An Application of Machine Learning Using Logistic Regression to Support Risk-Based Prioritization of Duty Drawback Claims under Section 29 of the Customs Act B.E. 2560 (2017)

Abstract

Duty drawback under Section 29 of the Customs Act B.E. 2560 (2017) is a key legal mechanism supporting national export competitiveness. However, in practice, customs authorities continue to face challenges arising from the increasing volume of drawback claims, data complexity, and limited human resources, which affect the efficiency and effectiveness of claim examination. This research aims to apply machine learning using a Logistic Regression model to support risk-based prioritization of duty drawback claims under Section 29, with the objective of enhancing risk management and decision support for customs officers. This study adopts a mixed methods research approach. Quantitative data were obtained from the information systems of Customs Department responsible for duty drawback administration, which constituted the research population, while selected samples were used for model development and evaluation. Qualitative data were derived from the analysis of legal documents, risk management frameworks, and relevant academic literature. The primary research instrument was a machine learning classification model based on Logistic Regression, developed to classify duty drawback claims into different risk levels.
Data collection was conducted through the extraction of administrative records from Customs Information Systems (CIS) and systematic document analysis. Statistical techniques employed in data analysis included descriptive statistics and performance evaluation metrics for classification models. The key findings indicate that the Logistic Regression model
can systematically support the prioritization of duty drawback claims by risk level,
enabling officers to focus their examination efforts on high-risk claims more effectively. The results demonstrate the potential of machine learning as a transparent and explainable decision-support tool for public sector risk management, while remaining consistent with legal principles, administrative fairness, and accountability in customs enforcement.