Big data, which I refer to as data-driven decision making , is used in each of the three functions of fraud prevention, i.e., automated screening, manual verifications, and chargeback management .
With automated screening, i.e., the process of testing payments automatically against whitelists, blacklists, force-review rules, and scoring algorithms, big data is used to perform the following tasks:
- identify the most important drivers of payment fraud ;
- calculate the relative importance of each risk factor ; and
- review the performance of all those models .
With manual verification, i.e., the process of manually reviewing payments for possible fraud, big data is used to perform the following tasks:
- explore custom risk reports ;
- make data-driven arbitrage decisions ; and
- audit the performance of manual verifications retrospectively .
Finally, with chargeback management, i.e., the process of reviewing who is liable for chargebacks (e.g., the issuing bank, the merchant, the cardholder, see “liability” ), big data is used to perform the following tasks:
- retrieve and review related cases of payment fraud; and
- build a defense for the chargeback case based on past cases.
[..] How is big data used to identify credit card frauds? Quora.
 Harvard Business Review on Making Smart Decisions (Harvard Business Review Paperback Series): Harvard Business Review: 9781422172391: Amazon.com: Books
 Chargeback - Wikipedia
 Feature selection - Wikipedia
 Calibration (statistics) - Wikipedia
 Model selection - Wikipedia
 Exploratory data analysis - Wikipedia
 Verification and validation of computer simulation models - Wikipedia
 EMV - Wikipedia