REVIEWER: Jong Dal Lee, CRM team manager for Kookmin Bank.

BACKGROUND: On November 1, 2001, the new Kookmin Bank was created through the merger of Kookmin Bank and H&CB. Ranked the 17th in Asia and the 68th in the world with approximately $147 billion in assets, the new Kookmin Bank is positioned as a world-class bank. The bank provides a full range of retail and commercial banking services to more than 14 million customers. It has a vast retail network consisting of 602 branches, 4,200 ATMs and a state-of-the-art Internet banking service which has 1.2 million online banking customers.

PLATFORMS: DW server (hardware): NCR WorldMark 4850 4 NODE; OS: UNIX MP-RAS 3.02; RDBMS: Teradata Database V2R4.0.

PROBLEM SOLVED: Teradata Warehouse Miner solved two important business challenges. First, a significant portion of the bank's revenue comes from overdraft protection for personal accounts. We used Teradata Warehouse Miner to identify those account holders with good credit ratings who had not yet set up overdraft protection. We approached them about setting up the new services, and this has helped us expand our relationships with existing customers and greatly boosted profitability. Second, to increase the sales of personal lines of credit, we use Teradata Warehouse Miner to identify those customers most likely to use this type of loan product. Then, through a rigorous analysis of customer banking behavior and credit ratings, we identified the most appropriate customers and made a loan offer with a reasonable credit limit.

PRODUCT FUNCTIONALITY: Once we installed our Teradata warehouse, Teradata Warehouse Miner enabled us to process vast amounts customer data quickly and cost-effectively without having to move it from the data warehouse. H&CB expects that the warehouse and mining tool will be able to handle the expected post-merger growth as well. In addition, the bank has used Warehouse Miner to apply one prediction model and develop an entirely new one.

STRENGTHS: Most data mining is done outside of the database, which is both expensive and time-consuming ­ and uses vital personnel. Working in conjunction with Teradata Data Warehouse, Teradata Warehouse Miner processes data directly within the database in parallel fashion, making it optimal for handling large volumes of data. In addition, because the two products are directly linked, it is easy to access, transform and process the data. Finally, Warehouse Miner saves costs and disk overhead because it eliminates the need for a separate mining server.

WEAKNESSES: For the first version of the product, Teradata Warehouse Miner provides a good base of functionality, but more is needed. Though it is better than most, like all data mining products, Teradata Warehouse Miner could use more statistical functions for data preprocessing. The current version only offers decision tree, clustering, factor analysis, linear regression and logistic regression.

SELECTION CRITERIA: Teradata Warehouse Miner's biggest advantage is its high price/performance ratio. In addition, we developed our mining model at the same time as we were implementing a CRM platform using Teradata Data Warehouse. Because we were already using Teradata Data Warehouse, Teradata Warehouse Miner allowed us to avoid the substantial costs of purchasing a separate mining server.

DELIVERABLES: Teradata Warehouse Miner provided the data preprocessing and analytical modeling techniques that enabled us to build three models and smoothly integrate them into our business process. We used both decision tree and logistic regression to build the Matured Deferred Deposit Churn and Load Account Prediction models. We also used Teradata Warehouse Miner's clustering algorithm to build a customer segmentation model based on customers' savings product preferences. These models are used to prevent customer attrition, increase services revenue and better serve our customers.

VENDOR SUPPORT: Teradata held regular and ad hoc meetings with business users to both review progress and plan ahead, allowing them to fully incorporate user requirements in the project. Even after the project was completed, Teradata continued to add value for our end users.

DOCUMENTATION: Teradata provided a user manual, meeting minutes, progress reports and outcome reports by mining area. The outcome reports included subject area definitions for each mining area, and definitions of populations, the data extraction method, the modeling method and a model evaluation. The documentation and training eventually made an unfamiliar product easy to use.

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