Company: Nationwide Building Society
Company background: A financial services mutual (owned by the members) providing banking services, with a focus on residential and private landlord mortgages, savings and current accounts in the United Kingdom. A branch network of over 550, serving 15m customers with around 16,000 employees, 600 active SAS users.
Contact: Philip Gee
Title: Senior SAS Data Consultant
Country: United Kingdom
Award Category: Innovative Problem Solver
Tell us about the business problem you were trying to solve?
When we have an Authorised Push Payment (APP) Fraud, we can’ t mark the fraudulent transactions on the account. We agree and record a Potential Loss amount. Then another team uses the information from the customer to find and mark the posted transactions include in the fraud. Can we automate this?
How did you use SAS to solve that business problem? What products did you use and how did you use them?
APP are often a few transactions. Can I test sets of transactions to match the Pot Loss? Using SAS 9.4 Enterprise Guide, data step ARRAY of transactions, processed by nested loops, tested r=5, n=125 (245m combinations) in 30s. Hash objects were used in later steps to find the best candidate set.
What were the results or outcomes?
A 70% success rate with exact matching, taking 10 mins at 0230 for about 70 cases, compared to a person taking 4 hours at 0800. Errors were due to badly recorded Pot Loss, more than 6 transactions in a case, too wide a date range (N=high) or other error. Run further cycles with relaxed constraints.
Why is this approach innovative?
Using the data in an array and nesting loops to test from 1 to 5 items, using macros to allow different r value was high performing. Hash objects for a non-sequential match of payee name to all other transactions to decide if these were new or previous payees and relaxing constraints was original.
What advice would you give to new SAS users?
Learn techniques so that you have options on how to solve problems. Think about what the problem is before you code. Simplify and then evolve to solve harder challenges. Engage with your customers to understand what they need. Develop a proof of concept quickly and progress what is successful.
