Background

A bank that issued credit cards to its customers was developing an analytical application to reduce the risk of credit card fraud by automatically monitoring purchase activity.  

Problem

The bank received transaction files from an outside organization monthly.  Each month the bank received a transaction header file with as many as 25 million records and a transaction detail file with up to 100 million records.  The bank sought to develop 13 month rolling averages to monitor transaction activity from month to month, as well as to the same month from the prior year.

Solution

Given the voluminous nature of the monthly source data files, and the fact that data must be aggregated across 13 consecutive time periods, the resulting migration solution had to accommodate massive volumes of data.  We selected C+ to pre-format the source data so that each data element conformed to a standard. 

Next we leveraged a third party data processing tool to rapidly filter the source data sets into smaller data sets that were relevant to the specific reporting requests which would be submitted. 

When these smaller files were formulated, they would be merged with similar data from the prior 12 months.  Once the final data sets were complete, Oracle Loader was used to expedite the load of the data into the database, where the reporting tool had immediate access to the data.

MicroStrategy was selected as the reporting tool because of its flexibility and ROLAP (Relational On-Line Analytical Processing) architecture.  Microstrategy provides a sophisticated algorithm for identifying the optimal data source to accurately satisfy a query in the least amount of time.

Benefits

Gave the Company Leverage to Ward off Fraud by Detecting Trends Before Too Much Damage was Done

Saved the nearly $200k cost of ETL Software and Related Implementation Costs by building the solution internally

Leveraged Cutting Edge Technologies to provide detailed graphical views of Credit Card Activity

Improved Data Quality Controls