Establishing risk and targeting profiles using data mining: Decision trees

Abstract

The application of technology and the computerisation of management processes in customs administrations have undoubtedly accelerated the processes related to data storage. In this context, customs administrations possess vast amounts of data on trade and financial flows. Data mining tools can be effective in analysing huge reams of data. Data mining consists of understanding, preparing, modelling and analysing data using different techniques, such as machine learning. Many of these techniques have advanced predictive analytical capacities that can ultimately lead to improved analytical capabilities in risk management. The Chi-square Automatic Interaction Detector (CHAID) decision tree method was selected for the purposes of this paper to determine the customs risk factors associated with import declarations recorded in the customs clearance system. The CHAID method is also used to create risk profiles and predict non-compliant customs declarations based on established rules.