Publications

Article published in DM Direct Newsletter - July 30, 2004 - Mike Meyer

Predictive Models Drive Today's Businesses

Just about every business has a need to predict or to understand specific aspects of consumer behavior. Financial institutions want to know about the risk of default or bankruptcy; cell phone operators need to understand attrition and churn; grocers want to know the likelihood that a particular discount offer will be accepted; advertisers want to know if a pitch will be successful. Modelers have created models across nearly all industry segments to help businesses make these predictions more accurately.

Frequently, predictions are based on pooled (across many businesses or even industries) models that use a minimum of common numerical attributes about the customer. When the problem is important and resources are available, modelers may build a custom model - one that predicts the exact outcome the business is interested in, using data specific to that business.

... In fact, using mixed data - data that incorporates structured, transactional and textual information - for predicting customer behavior has proven to increase model accuracy routinely by more than 20 percent across a wide range of modeling problems.

Here is the rest of the story.