[WEBINAR] The New Normal: Dealing with the Reality of an Unsecure World

Real-Time Fraud Detection

Definition - What does Real-Time Fraud Detection mean?

Real-time fraud detection is the real-time execution of fraud-detection algorithms in order to detect fraudulent activities on credit cards and other financial payment systems. It makes use of real-time data analysis such as forensic analytics and predictive analytics to determine if an ongoing transaction is legitimate or not. Though the system is not perfect, it has reduced fraud losses in the U.S. by 70 percent since 1992, when real-time fraud detection was introduced.

Techopedia explains Real-Time Fraud Detection

Fraud detection in the simplest form is simply outlier detection, which is determining whether an event such as a purchase using a credit card occurs outside of the normal circumstances or habits of the person using it. Real-time fraud detection is just the execution of fraud detection algorithms right as the purchase is happening. The system is not perfect and a lot of false positives are captured, but this just ensures that fraud is detected immediately and possibly prevented outright. For example, a man that has exclusively been using his credit card to purchase gadgets online suddenly purchases women’s lingerie in a store from a town far away from his home. This would immediately register as an outlier occurrence because it deviates so much from the person’s purchasing habits, and depending on the credit card issuer, the transaction might be blocked or the person would get a call immediately afterwards from a representative in order to confirm whether the recent purchase was legitimate or not.

Before real-time systems made fraud detection instant, it used to be done in bulk with the results often arriving weeks or months after the purchase, which makes it difficult to track down the fraud or has allowed the culprit to commit many more fraudulent purchases before being detected and caught. This was because data used to be stored on slower disks since memory was still relatively costly. But since the cost of memory has gone down significantly since the early '90s, it has become possible to store data in-memory so that processing can happen very quickly. Real-time fraud detection can occur in as little as 40-60 milliseconds; in comparison, a human eye blink happens in 300 milliseconds. As of today, real-time fraud detection is a very common use case in the field of big data.

Techopedia Deals

Connect with us

Techopedia on Linkedin
Techopedia on Linkedin
"Techopedia" on Twitter

Sign up for Techopedia's Free Newsletter!

Email Newsletter

Join thousands of others with our weekly newsletter

Free Whitepaper: The Path to Hybrid Cloud
Free Whitepaper: The Path to Hybrid Cloud:
The Path to Hybrid Cloud: Intelligent Bursting To Amazon Web Services & Microsoft Azure
Free E-Book: Public Cloud Guide
Free E-Book: Public Cloud Guide:
This white paper is for leaders of Operations, Engineering, or Infrastructure teams who are creating or executing an IT roadmap.
Free Tool: Virtual Health Monitor
Free Tool: Virtual Health Monitor:
Virtual Health Monitor is a free virtualization monitoring and reporting tool for VMware, Hyper-V, RHEV, and XenServer environments.
Free 30 Day Trial – Turbonomic
Free 30 Day Trial – Turbonomic:
Turbonomic delivers an autonomic platform where virtual and cloud environments self-manage in real-time to assure application performance.