Lancaster University Management School Uses SAS® Analytics Software to Help Online Retail Company Predict Fraud Risk
Lancaster University Management School’s Department of Management Science is using SAS® software on a collaborative project with a leading online and home shopping retailer in the UK. Masters students use the software to provide research and consulting services to help predict fraud risk associated with Goods Lost in Transit (GLIT) - a common issue experienced by the online and home shopping sector.
Lancaster University Management School (LUMS) is a triple-accredited world-ranked management school. The school's Department of Management Science has an outstanding research record. In rolling out masters courses, it engages with industry partners that help provide students with academic and commercially focused work. SAS supports this approach through its academic programme. It does this by developing courses on which SAS® software is used, and by supporting the placement of students as interns at various commercial organisations where they work on projects using SAS software to solve industry challenges.
The department has recently been using the latest in SAS® Analytics software to help students on a masters dissertation project address issues around Goods Lost in Transit (GLIT) faced by one of the leading online and home shopping retailers in the UK.
Every year, this retailer suffers millions of pounds in losses through false claims of GLIT. To prevent the fraud claims, the retailer can ask for a signature on delivery as a proof of delivery (POD). However, doing this on every delivery increases costs, and potentially leads to cancelled orders and lost revenues.
To address this issue, an internal consulting team was commissioned to build an analytical tool to predict the fraud risk (GLIT rate) associated with each order, product and customer. This allows it to identify on which deliveries it would be most valuable to ask for signed deliveries. It turned to the LUMS' Department of Management Science and its masters students for consultancy on the process. Over the years, the department has been working with this retailer in tackling all sorts of challenging business issues.
Using SAS® Enterprise Miner™
LUMS decided to use SAS® Enterprise Miner™ to build the requisite scorecard model (logistics regression model). There were three main reasons for this. First, SAS Enterprise Miner is a powerful analytics tool. Second, it is well-integrated with the retailer's database system and the retailer has a very strong analytical capability and, finally, it has been widely taught on the department's SAS programming and data mining courses.
SAS Enterprise Miner was the key tool in addressing the three main technical challenges presented by this project. First, while extracting data from the company’s online database was relatively easy, decoding, cleaning and processing data have proved a more protracted process. However, SAS Enterprise Miner can be tightly integrated with the company’s database. As a result, SAS users can reduce time taken to carry out these tasks.
Second, the SAS software is used to overcome the technical challenge posed by sampling. While losses can be high due to the large volume of transactions, the GLIT rate is very small (normally less than 1%), which causes insignificant statistical tests. Therefore, under simple random sampling strategy, the selected predictors are insignificant. To enhance the predictability, the project team did some rigorous analysis and decided to select a sample, split 50/50 between GLIT and non-GLIT cases and analysed this sample using SAS Enterprise Miner. The treatment helps obtain desired statistical significance. They then transferred the adjusted GLIT scores to the actual scores.
The third challenge is to use modelling to identify factors affecting the GLIT rate. The team used SAS Enterprise Miner to analyse the relationship between the GLIT rate and a range of variables including addresses and postcode locations, and customers' historical purchase information. Finally, the project team also used the software to estimate the probability that, having been asked to sign on delivery, customers would cancel orders.
The pooling of resources between the retailer and LUMS' Department of Management Science has succeeded in delivering advantage to both groups.
According to Dr Zhan Pang, lecturer in operational research, LUMS, "The collaboration between this industry leader and LUMS is mutually beneficial. On the one hand, it provides students with the opportunity to practise the knowledge and skills they have learnt on campus in a real commercial context. On the other, it gives the company more research capability and consulting service provided by masters students and academic supervisors."
The use of SAS Enterprise Miner on the project also proved to be a success. The product streamlines data mining to create accurate predictive and descriptive models based on large volumes of data from across the enterprise. This enables it quickly and efficiently to decode, clean and process a significant quantity of data contained within the company's online database.
According to Dr Pang, "Fraud GLIT claims are threatening the UK's fast-growing online retailing business, costing about £48.5 million per year according to market research by Interactive Media in Retail Group (IMRG). Preventing frauds taking place across millions of customer orders in real time is extremely challenging and demands advanced analytical modelling techniques and data analysis tools. SAS Enterprise Miner is an excellent data mining tool which can leverage large volumes of enterprise data to create accurate predictive and descriptive models for fraud detection.
The results illustrated in this article are specific to the particular situations, business models, data input, and computing environments described herein. Each SAS customer’s experience is unique based on business and technical variables and all statements must be considered non-typical. Actual savings, results, and performance characteristics will vary depending on individual customer configurations and conditions. SAS does not guarantee or represent that every customer will achieve similar results. The only warranties for SAS products and services are those that are set forth in the express warranty statements in the written agreement for such products and services. Nothing herein should be construed as constituting an additional warranty. Customers have shared their successes with SAS as part of an agreed-upon contractual exchange or project success summarization following a successful implementation of SAS software. Brand and product names are trademarks of their respective companies.
Copyright © SAS Institute Inc. All Rights Reserved.
Lancaster University (Fraud)
Create accurate, predictive and descriptive models for detecting fraud claims of Goods Lost in Transit (GLIT). Identify customers and products best suited to an approach based on signed proof of delivery (POD).
SAS® Enterprise Miner™ used for decoding, cleaning and processing data, for modelling and to estimate the probability that customers would cancel orders.
Project identified eleven factors that influence the GLIT probability significantly. Management report provided, reporting project delivery process, insights gained and recommendations to develop fraud risk management strategy. Project team found that even targeting less than 20% of all orders they could still capture 25% of GLIT cases, potentially saving the business millions of pounds.
“Fraud GLIT claims are threatening the UK's fast-growing online retailing business, costing about £48.5 million per year according to market research by Interactive Media in Retail Group (IMRG). Preventing frauds from taking place across millions of customer orders in real time is extremely challenging and demands advanced analytical modelling techniques and data analysis tools. SAS Enterprise Miner is an excellent data mining and analytical modelling tool which can leverage large volumes of enterprise data to create accurate predictive and descriptive models for fraud detection.”
Dr Zhan Pang
Dr Zhan Pang, lecturer in operational research, Lancaster University Management School