Naive bayes algorithm in hs code classification for optimizing customs revenue and mitigation of potential restitution

The Directorate general of customs and excise, as a government revenue collector, must maximise import duty receipts each year. One common issue is the return of unpaid import duty and/or administrative punishments in the form of fines based on the objection judgement document. The tax court could help you minimise your gross receipts at the customs office. Data mining techniques are intended to provide valuable information regarding the HS Code classification technique, which can assist customs agents in determining duties and/or customs values. This study makes use of data from the notification of Import of Goods at customs regional office XYZ from 2018 to 2020. The Cross industry standard process for data mining (CRISP-DM) model is used in this study, and the naive bayes algorithm in rapidminer 9.10 is used for data classification. According to the model, the calculation accuracy is 99.97 percent, the classification error value is 0.03 percent, and the Kappa coefficient is 0.999.