SPSS - UK Customers

ECtel Ltd

Data mining powers telecommunications fraud detection

The situation

ECtel Ltd, located in Israel, is a leading provider of fraud detection and prevention software for wireline, wireless, and next generation networks.

The challenge

There are many ways to avoid paying for telecommunications services, from stealing phone card numbers to bypassing phone circuitry. The result? Fraudulent activity costs the industry billions per year, forcing some companies to go out of business.

The solution

ECtel created FraudView™. a comprehensive fraud management solution. The solution uses SPSS Inc.'s advanced data mining workbench, IBM SPSS Modeler*, to support FraudView's MineView engine, which is used for detecting telecommunications fraud in real-time.

Selling a data mining product to a telecom provider has been difficult, traditionally, because they don't have data mining experts on staff who can work conventional data mining tools. IBM SPSS Modeler's functionality allowed ECtel to develop software based on IBM SPSS Modeler as the internal data mining engine.
Eric Kaplan
FraudView group manager
ECtel Ltd

The results

Most people don't rent an apartment with the intent of setting up a switch-board to make free phone calls. Unfortunately, some unscrupulous individuals do, cheating telecommunications companies out of billions of dollars in revenue. In Israel alone, this type of fraudulent activity costs service providers an estimated $150 million each year, according to one telecommunications consulting firm.

Since no international organisation monitors fraud, telecom providers are left to deal with these crimes on their own. To address this challenge, ECtel Ltd introduced FraudView.

FraudView is a fraud protection system that gathers information about calls in real time, and compares them against fraudulent scenarios produced by its MineView component. MineView uses SPSS Inc.'s IBM SPSS Modeler to evaluate a sample of an ECtel customer's call records, and then generates scenarios that most accurately predict fraud. If MineView flags the subscriber as fraudulent, an alert is automatically generated and filed into a case profile for the telecom investigator to examine.

Significantly reduced telecommunications fraud for more than 150 companies worldwide

More than 150 blue-chip telecom service providers worldwide, including national carriers in Germany, France, China, and the United States, have turned to ECtel's FraudView to manage fraud. According to Eric Kaplan, FraudView group manager for ECtel, "Selling a data mining product to a telecom provider has been difficult, traditionally, because they don't have data mining experts on staff who can work with conventional data mining tools. Its functionality allowed ECtel to develop software based on IBM SPSS Modeler as the internal data mining engine".

Another convenient feature that telecommunications investigators appreciate is what Kaplan refers to as ‘data mining for the people’. MineView automatically self-calibrates its fraud scenarios, adjusting them according to recent call detail entries, the fraud officer's feedback, and information from external sources like the billing system. "By functioning autonomously, MineView allows non-data mining people to operate a data mining workbench," said Kaplan.

Saving money by enabling real-time fraud detection

For each subscriber, MineView calculates a suspicion score – the probability that a fraudulent subscriber is using a suspect number – based on data collected in real time. Whenever this score exceeds the preset threshold, MineView sends an alert notifying the investigator of a suspicious subscriber. This real-time data collection is a significant advantage over other data mining systems, which compare data against the model at two-hour intervals.

"Upon seeing a single alert from MineView, the investigator is ready when a second call appears from that subscriber, and can block that call as it's happening," explained Kaplan. Thus money is saved because fraud is detected and acted upon immediately.

As new telecommunications service providers continue to enter the worldwide market, the demand for sophisticated technology that identifies and protects against fraudulent subscribers will increase significantly. With SPSS Inc.'s IBM SPSS Modeler operating as its real-time fraud detection engine, ECtel's FraudView offers the telecommunications industry a proven fraud management solution.

Interested in how predictive analytics can detect and prevent fraud? Download the ECtel Ltd PDF here.

*IBM SPSS Modeler, formerly called Clementine®, is part of SPSS Inc.’s Predictive Analytics Software portfolio.

Customer Success Stories