Analytic Techniques: Cross-Tabulation  Use the index below to go directly to a section or use the back/next buttons Cross-tabulation         Simple Profiling   Regression CHAID Other Techniques Data Mining Applications Typical Data Mining Process

Cross-tabulation

Cross-tabulation is the simplest technique for understanding patterns of differences between populations in a database. The previous bar graph showing patterns of age distribution for both customers and non-customers is actually a graphical representation of some information taken from a cross-tabulation of age and customer status.

Thus, cross-tabulations give us much more insight into the data than do simple profiles or frequency distributions. Cross-tabulations are an example of bivariate analysis (i.e., examining the relationship between two variables).

However, cross-tabulations are of limited value, too, because we are realistically restricted to examining the relationship between only two variables at a time (e.g., customer status and age; customer status and income; etc.). If we try to examine a cross-tabulation of more than two variables at a time, the results are very difficult to visualize and interpret meaningfully. (Later on, we will discuss more advanced techniques that help us solve this problem.)

Of course, we can run several separate cross-tabulations and then try to combine the results to get a final answer. For example, we can run separate cross-tabulations of customer status and age; customer status and income; etc. And then we can try to figure out a target definition based on all these separate cross-tabulations.

However, trying to piece together separate cross-tab analyses to form an overall picture can be both difficult and misleading. For example, if we examine cross-tabs for customer status by age, customer status by income, etc., we still do not know the relative importance of each demographic in discriminating between customers and non-customers. To do this, we need to turn to multivariate techniques, which examine the relationships between more than two variables at a time.

	Cross-tabulation
Simple Profiling		Regression	CHAID	Other Techniques	Data Mining Applications	Typical Data Mining Process