Logistic regression is a technique for modeling and predicting categorical data. Although it is a standard technique in biomedical research, where researchers routinely deal with the relative risk of the outcome in one group compared with another group, most people outside of this field fail to use it when they should – probably due to lack of training or stats-averseness.
 However, there is no dearth of attribute processes in the transactional arena! Master Black Belts and Black Belts would do well to understand the mechanisms and principles associated with binary logistic regression (as they do differ in many respects from continuous variable multiple regression) and how they extend to multinomial and ordinal variables.
Join Smita Skrivanek, Principal Statistician at MoreSteam.com, in this recorded Webcast as she discusses how to use and teach the concepts of binary logistic regression. This session includes:
- What is binary logistic regression
- When a logistic approach is appropriate
- The maximum likelihood method of estimating model coefficients
- How to evaluate model performance or comparing alternative models
- How to interpret residuals and some potential problems and their solutions.
Recommend This Webcast to Someone Else
Send an email to a colleague to let him or her know about this Webcast!
Other Webcasts in the 2010 MBB Summer Webcast Series
|
