Publicação
Enhancing merchants' identification for payment systems using the optics clustering algorithm and similarity score
| Resumo: | Inconsistent merchant names in payment transactions present a significant challenge for a European payment processor in its daily operations, leading to incorrect labeling of merchants in transaction records. This thesis develops an unsupervised model that identifies the most frequent merchants from transaction data and integrates these into an existing merchant labeling model. The approach uses text cleaning, similarity measures, dimensionality reduction, and clustering (OPTICS) to group similar names. Results show this integration improves labeled transaction coverage at least in 3%, reduces manual effort, and ultimately strengthens the company’s virtual card services. |
|---|---|
| Autores principais: | Ferreira, António Carlos Fiúza |
| Assunto: | Merchant identification String similarities Merchant profiling OPTICS Clustering SparcePCA Payment systems Business analytics |
| Ano: | 2025 |
| País: | Portugal |
| Tipo de documento: | dissertação de mestrado |
| Tipo de acesso: | acesso aberto |
| Instituição associada: | Universidade Nova de Lisboa |
| Idioma: | inglês |
| Origem: | Repositório Institucional da UNL |
| Resumo: | Inconsistent merchant names in payment transactions present a significant challenge for a European payment processor in its daily operations, leading to incorrect labeling of merchants in transaction records. This thesis develops an unsupervised model that identifies the most frequent merchants from transaction data and integrates these into an existing merchant labeling model. The approach uses text cleaning, similarity measures, dimensionality reduction, and clustering (OPTICS) to group similar names. Results show this integration improves labeled transaction coverage at least in 3%, reduces manual effort, and ultimately strengthens the company’s virtual card services. |
|---|