Posts Tagged ‘fraud detection’

Build Fraud Detection Model Without Historically Known Fraud Data

Tuesday, May 22nd, 2012

When there are no historically known fraud data, we have to build unsupervised models that do not require a target variable. There are a number of such models that we can try: one-class support vector machine, principal component based compression net, etc. They are detecting data points that are abnormal, not fraudulent per se. We only hope that  being abnormal is highly related to being fraudulent.

Visa, MasterCard affected by ‘massive’ data breach

Saturday, March 31st, 2012

On the other hand,  small data breach events (e.g, skimming) are happening on daily basis (a few hundred or thousand cards are skimmed here and there across millions of POS/ATM). They are less dramatic and wouldn’t make the headline. But their cumulative damage over a year is huge. That’s why we have developed Fraudulent Card Early Detection Solution.

Long Term Card Fraud Prediction

Sunday, October 30th, 2011

Most of the debit/credit card fraud detection is based on transaction level rules or scores that target the fraud transactions happening in real time. Fraud prevention operation is struggling to stop fraud in real/near real time.

If we can predict which cards will likely have fraudulent activities within the next a few months, we can either reissue those cards today or build transaction level rules to monitor them. The long term card fraud prediction is, just like long term weather forecast, doable.  At BDM, we have tested such a method and found the results are good and actionable.

“Superficial” vs. “Inherent” Fraud Detection Rules

Tuesday, August 30th, 2011

There are two types of rules, “superficial” rules and “inherent” rules. Superficial rules are easier to create and they are straightforward to understand, e.g., transactions happening at certain retailers. These rules may be good for a few weeks and their performance drop quickly. Inherent rules capture more time lasting ( and usually hidden and mutli-dimensional) patterns of fraud behaviors. To create them,we may have to use more sophisticated methods including predictive modeling. The PRM SQL statements for some of these inherent rules that we created have 5,000 characters. These rules work like magic and have very low false positives.

BDM’s Debit Card Fraud Detection Rules Won A Contract With One Of Top 15 Banks

Monday, September 27th, 2010

BDM signed a contract with one of the top 15 banks in America to provide debit card fraud detection rules. In a head to head competition, BDM’s rules detected far more frauds given the same number of alerts based on historical data than competitors did.  BDM rules were deployed into production and have screened almost 200 million debit card transactions in real time. They have delivered solid results and achieved annual ROI of 1,000%.   BDM rules are unique in that they detect fraud with extremely high accuracy statistically while they are interpretable by fraud prevention staff.

Early Fraud Detection System

Monday, March 22nd, 2010

Traditional fraud prevention methods try to detect fraud when it happens. By exploring the unusually behavior patterns long before when fraud occurs, we are able to proactively prevent fraud with minimum impact to good customers.

Finding Fraud Detection Rules Through Evolution!

Monday, January 18th, 2010

Through the principle of  survival of the fittest, the natural evolution can find the best genes for the environment. A genetic algorithm (GA) simulates the natural evolution process to search the best solutions or fraud detection rules in our case. With our cutting edge GA based proprietary technology, a large number of initial fraud detection rules evolve to detect more fraud at lower false positive. After many generations of evolutions, the best fraud detection rules are the ones that survive.

The true story of a bank card holder.

Friday, January 8th, 2010

It is a true story. I met David at Predictive Analytics World in Washington DC last October (of course, he later became one of my linkedin connections). He told me emotionally that in the past a few months his bank card transactions were blocked twice by the bank’s fraud prevention. He said he had stopped using the card and would cancel it in a few months.

I realize how serious the situation is for the bank.

Firstly, the bank will lose the future income from David if he was not bothered by the bank’s fraud prevention and might stay with the bank for, say, 5 more years. Supposing the bank makes $300 every year from David, that would be equivalent of $1,400 loss in present value.

Secondly, we can say that a lot of the time the fraud prevention is making mistakes. If the fraud prevention places holds on 100,000 cards a year, how many of them are mistakes and thus inconvenience good customers like David? 10,000? or 50,000? or 80,000? or 99,000?

Fraud prevention is a double-edged sword. On the one hand, it saves fraud losses. On the other hand, it reduces good revenue which could be significant. Do you want to share your insights on this?

If you are a C level executive or manager who owns the whole card or account operation, it is the bottom line that matters. I developed a methodology that uses the notion of the total cost to measure the impact of a fraud prevention to the bottom line. The total cost includes fraud loss, operation cost and opportunity loss due to false positive. The approach is not perfect but still useful. We can not afford not to measure the cost of inconveniencing good customers by fraud prevention. Please send me an email ( if you want the document.