Teradata Tool Digs Up Customer Info

Teradata this week will announce the next version of its Teradata Warehouse suite, which has additional data mining capabilities that can help companies better predict customer behavior.

Teradata this week will announce the next version of its Teradata Warehouse suite, which has additional data mining capabilities that can help companies better predict customer behavior.

Teradata, a division of NCR Corp., of Dayton, Ohio, will introduce Teradata Warehouse 6.2, which includes Warehouse Miner 3.2, at its Teradata Universe conference in Edinburgh, Scotland. The new Warehouse Miner will include two additional algorithms that support affinity analysis and sequential analysis.

Both algorithms help companies—particularly those in retail, telecommunications and financial services—plan product offerings, promotions and ways of preventing the loss of customers, according to Teradata officials, in San Diego. Warehouse Miner already supports eight other algorithms, such as linear regression and factor analysis.

Affinity analysis enables companies to better plan their product offerings by predicting which products customers will purchase together. Sequential analysis relates two or more events over time so that companies can predict which products customers are likely to buy in their next transactions.

Teradata first began integrating data mining into its database engine in 1999 with pre-processing capabilities. Last April, the company launched a more comprehensive data mining tool with analytic algorithms with the first release of Warehouse Miner.

The additional algorithms Teradata is supporting are among the most popular, especially for retailers and telecommunications companies, according to Evan Levy, an analyst at Baseline Consulting Group, in Los Angeles.

The algorithms support in the database engine will especially benefit those analyzing terabytes of data that can span hundreds of thousands, or even billions, of records. Users wont have to offload the data for the analysis and can take advantage of the parallel processing inherent in Teradata, making data mining simpler and faster, Levy said.

"Theyre removing a very time-consumed step by embedding data mining into the engine itself," Levy said. "Much of the effort in data mining is walking through the data and comparing the data. When you break it into parallel steps, theres a dramatic improvement in getting the job done."