Will anybody buy? Logistic regression - Improving the User Experience through Practical Data Analytics - page 295

FIGURE 11.16
Output for CharlestonGlobe.com binary logistic regression; SPSS.
SIDEBAR: HOW HIGH IS HIGH?
It is dificult, if near impossible, to say whether you should consider 70% to be a high or not-so-
high value of an r 2 , since that issue is so context dependent, and there is little previous work in
the public domain to compare the 70% to. As we noted in the previous two chapters, there are
situations when 70% would be viewed as fabulous (e.g., trying to predict who will buy and who
will not buy in a catalog marketing endeavor), while in other situations when it clearly is not high
enough to use the results to strong advantage. Still, the authors' instinct in this case is that it is
decently high.