Customs Import Declaration Datasets

本数据集由韩国科学技术院创建，包含54,000条人工合成的进口申报数据，涉及22个关键属性。数据集通过条件表格GAN合成，保持了特征间的相关性。该数据集解决了原始进口数据无法公开的问题，同时保持了与源数据相似的分布，适用于多种下游任务，如分类算法性能测试。此外，数据集还被用于教育目的，如世界海关组织的高级数据分析课程，以及大学间的欺诈检测算法竞赛，旨在提升数据分析技能，促进海关领域的研究与创新。

This dataset was developed by the Korea Advanced Institute of Science and Technology (KAIST). It encompasses 54,000 synthetic import declaration records with 22 key attributes. Generated via Conditional Tabular GAN, the dataset preserves the correlations between features. This work addresses the challenge of non-publicly available original import data, while maintaining a distribution similar to that of the source dataset, making it applicable to various downstream tasks such as classification algorithm performance testing. Additionally, the dataset has been utilized for educational purposes, including advanced data analysis courses offered by the World Customs Organization (WCO) and inter-university fraud detection algorithm competitions, aiming to enhance data analysis skills and promote research and innovation in the customs field.