On using data science to detect frauds

On 23 June, HasGeek in association with WalmartLabs organised a meetup on the role of data science in detecting frauds. The aim of the panel was to map out the landscape of frauds in transactions, identify the known and unknown types of frauds and the techniques used in data science to detect such frauds.

The discussion was moderated by Vinayak Hegde, CTO Zoomcar who spoke about his experience in helping build systems which would flag users who were conducting fraudulent activities. The panelists, Vivek Mehta, Nirmal J M and Vamsi Varanasi added inputs based on their experience in dealing with frauds.

Among the topics covered, the important ones were around how there was an increasing need to use advanced techniques and analytics to identify fraudulent activities. As it has become easier to buy products and avail services online, many people tend to misuse the convenience to defraud the companies offering such services. With the scale of these companies ever expanding, it is difficult to flag fraudulent transactions.

Further, with e-commerce companies such as Flipkart and Amazon, such frauds are increasingly getting sophisticated. The panel discussed further on why an organisation needs to have systems in place to prevent, detect and investigate fraud, this includes the Visa & MasterCard guidelines, government regulations on the fraud prevention and, importance of policies for safeguarding both merchants and customers.

Nirmal, one of the panelists for this discussion helped summarise the discussions at the meetup.