BACKGROUND: With assets of $205 billion, the Bank of Montreal is one of the 10 largest banks in North America.

PLATFORMS: RS/6000 hardware with AIX operating system.

PROBLEM SOLVED: Understanding the needs of our 18 million plus customers and analyzing the profitability of a rapidly changing product and service portfolio was, until recently, virtually impossible. Now, a data warehouse which we call the customer knowledge database (CKDB) is not only making such analysis possible, but also yielding rich dividends. With the kind of information that is now available to various divisions in the bank, such as the credit card division, loan division and the online banking arm, we have a better understanding of our customers' financial behavior and how it impacts our profitability. For instance, we have been able to segment our client base for more targeted marketing, refine products and services, price them suitably, design rewards programs and help fix interest charges.

PRODUCT FUNCTIONALITY: As the use of DB2 Intelligent Miner from IBM increases, we anticipate that growth will be accommodated either by moving toward a massively parallel processing environment or by installing DB2 Intelligent Miner on more client workstations. The product's performance and flexibility make both options feasible.

STRENGTHS: Having scouted around the market, we were not able to find any other data mining tool as scalable as DB2 Intelligent Miner. Also, because of DB2 Intelligent Miner's support for parallel processing, query time is fast. When you have large databases like our customer knowledge database, where you have more than 18 million accounts, you appreciate the speed and scalability.

WEAKNESSES: The meta data behind Intelligent Miner exists only in proprietary data. As this is not a very well-structured format, integration with other data mining tools is cumbersome. Also, the previous versions of Intelligent Miner are missing lift charts, which make it difficult to determine the best algorithm. However, IBM has corrected this weakness in the most recent version of the product.

SELECTION CRITERIA: The beauty of DB2 Intelligent Miner is that it can process huge volumes of data. The market is flooded with neural networks on PCs. However, you cannot run large volumes of data on such systems, so you have to use the old-fashioned sampling method ­ which does not always yield the most accurate results. With DB2 Intelligent Miner, we can mine as large a volume of data as necessary. In addition, Intelligent Miner's large number of reliable algorithms influenced our decision to select the product. And as our data resides on IBM DB2 Universal Database, Intelligent Miner was the logical choice for compatibility.

DELIVERABLES: Intelligent Miner produces segmentation models for the bank, such as the propensity of our customers to buy related products as well as the probability of customers to leave their households.

VENDOR SUPPORT: IBM's total-solution approach has ensured that the bank is a satisfied customer. As always, IBM has provided exemplary support, such as access to the IBM labs in Dallas, Austin and Germany. Both on the phone and in person, IBM has helped every step of the way.

DOCUMENTATION: One shortfall of Intelligent Miner's documentation is that it lacks detailed descriptions of error codes. Also, it would be beneficial to us if the algorithms were described in more detail.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access