News
PayPlug announces beta launch of its fraud prediction tool
November 2015 by Marc Jacob
PayPlug, which received a record 1.75 million euro subsidy earlier this year as part of the European Commission’s Horizon 2020 initiative, announces the beta launch of its next generation fraud-prediction tool. Unlike traditional rules-based fraud-prediction models, PayPlug’s technology uses predictive analysis and machine learning. This approach targets fraud attempts more precisely, adapts automatically to new fraud strategies, and bypasses the need to continually adjust parameters, which is part and parcel of rules-based models.
How Data Science will revolutionize fraud prevention?
The growth of e-commerce in Europe has brought with it an increase in online payment fraud . In order to protect themselves, merchants have historically looked to the same small group of major counter-fraud service providers such as Cybersource or Accertify. Their rules-based counter-fraud solutions require a merchant to continuously adjust a large set of parameters, which will then determine whether a transaction should be considered risky or not.
Breaking with these traditional models, PayPlug decided to incorporate machine learning into the development of its fraud-prediction tool. By ‘teaching’ the predictive models with behavioural analysis of online consumers across a high volume of transactions, the tools developed by PayPlug are able to predict the risk level of a new payment in real time.
After a year of research and development, PayPlug has announced today the launch of a beta version of their fraud prediction API. This will allow online merchants, as well as other payment service providers and banks, to enhance their counter-fraud efforts by using PayPlug’s fully automated, adaptive, and precise fraud-prediction tool.
In order to feed its predictive models, PayPlug is particularly interested in analysing the behaviour of internet users, specifically the intricacies of keystrokes and mouse movements.
