Breytenbach , JohanTemmers , Shaun2026-10-232026-10-232026https://hdl.handle.net/10566/25717While focusing on import product classification as it pertains to retailers in South Africa, this study considered how machine learning could redefine retail efficiency by automating the classification process as relates to imported products. The automated classification of a product using South African import compliance taxonomy is novel. The three dimensions of classification relevant to this study were HS Code, Compliance and Trade Agreement classification. As international sourcing has grown exponentially, the influence of the import supply chain has increased significantly and contributed to retailers being more competitive in their local markets. Thus, managing import product ranges and the relevant information systems meant to support this function has become more important than ever. Yet for many organisations classifying products remains a manual task with little system support and much reliance on human expert knowledge. To complete this study, import compliance classification data was collected from a South African retailer. This data was cleaned and statistically analysed and used to train various supervised machine learning models.enInternational trade complianceProduct complianceCustomsFree Trade AgreementArtificial intelligenceThesis