topic Question about the Spearman-Hoeffding Ranks interpretion in SAS Academy for Data Science

Question about the Spearman-Hoeffding Ranks interpretion

WWD — Thu, 09 May 2019 15:52:51 GMT

In the Module 4, chapter “Predictive Modeling Using Logistic Regression” course notes, the author makes a statement about the non-linear relationship between a predictor variable and the logodds dependent variable.

On page 3-49 of the notes, the author says that DDABal has a low Spearman value and a high Hoeffding value. On page 3-47, for DDABal, Spearman’s rank = 23 and Hoeffding’s Rank = 3.

What output is the author looking at to base their statement about the respective ranks for DDABal?

Thank you,

Bill Donaldson

Re: Question about the Spearman-Hoeffding Ranks interpretion

Cynthia_sas — Thu, 09 May 2019 16:09:17 GMT

Hi:
I've sent your question to the course instructors for follow-up.
Cynthia

Re: Question about the Spearman-Hoeffding Ranks interpretion

TWoodfield — Thu, 09 May 2019 16:58:23 GMT

The Hoeffding rank of 3 indicates that only two other variables have a larger Hoeffding value, namely SavBal and Dep. Thus, it is the third highest value, so it seems reasonable to say it has a high Hoeffding value. The Spearman rank of 23 implies 22 out of the remaining 36 variables have a higher Spearman value, so it seems reasonable to say that DDABal has a low Spearman value, although it is true that 14 variables have a lower value. A low rank implies a high value. The actual values are shown in the table on page 3-47 along with the ranks.

I hope I am not misunderstanding your question.

Terry Woodfield