A carregar...
Publicação
How to predict if direct debit payments bounce - a data analysis at one of Europe-s leading investment platforms
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 1.39 MB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
At one of Europe’s leading investment platforms, almost half of paid-in capital is transferred
via direct debit payment. Despite its large positive business impact, it entails the risk of a
payment bouncing. This paper analyses the platform’s direct debit usage and relates it to the
risk of bounced payments. Given the platform’s reactive risk approach, a proactive bounced
payment prediction model using anomaly detection is suggested. This correctly predicts 98
percent of bounced payments with an error margin of one in three payments. Deployment in
production would eliminate manual effort of tracking and following up on clients with bounced
payments.
Descrição
Palavras-chave
Business analytics Data analytics Machine learning Binary classification Outlier detection Fintech Online Brokerage Wealth Management Direct debit payments Risk management