Data defines the model by dint of genetic programming, producing the best decile table.


Modeling a Skewed Distribution with Many Zero Values
Bruce Ratner, Ph.D.

The standard approach for modeling a continuous target variable is the ordinary least-squares (OLS) regression model. One of the assumptions of OLS regression model is that the target variable is nonskewed, continuous with permissible discontinuities and minor clumping at several values, including the value zero. The OLS regression model for a skewed distribution with many zeros is not appropriate, and would assuredly render questionable results, or the floccinaucinihilipilification (estimating as worthless) of the analysis itself. The purpose of this article is to present a befitting method - the GenIQ Model© - for modeling such a distribution, regardless whether the zeros are indeed the lowest values or censured values, a situation quite common in database marketing. I illustrate the modeling approach using direct marketing data.
 
For more information about this article, call Bruce Ratner at 516.791.3544 or 1 800 DM STAT-1; or e-mail at br@dmstat1.com.
Sign-up for a free GenIQ webcast: Click here.