turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

Find a Community

- Home
- /
- Analytics
- /
- Stat Procs
- /
- Proc Logistic or Proc PHreg?

Topic Options

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

09-05-2011 05:07 AM

Hi all,

Could anyone explain to me what is the difference between the above mentioned methods? Trying to build a model for churn and pondering over which one to choose to achieve better results

Thnx in advance,

Aristos

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

09-06-2011 02:23 AM

The most differenct is data.

proc phreg is used to test influence of independent variables for censored data.

If your data is censored ,then use proc phreg.

proc logistic is used to test binary dependent variable.

Ksharp

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

09-06-2011 03:44 AM

Thank you for your clarification Ksharp, more clearer now what to use.

KR

Aristos

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

09-06-2011 10:13 AM

Logistic regresson in general deals with situations where you are trying to classify subjects into cases, often binary, but it may be ordinal (low, medium, high), or nominal (choice of food, for example, there is no scale or ranking within these classes). SAS has a number of procedures that can handle logistic regression, PROC LOGISTIC being one of them, and has been around a long time. But it's not the only PROC that can handle logistic regression problems.

PHREG - proportional hazard models deal with survival analysis where the TIME dimension is very prominent in the analysis. The interest is on the hazard function. The data structure is quite a bit more complicated, especially if youhave time-varying covariates.

For churn problems, sometimes for convenience, one picks a fixed timeframe, say 6 month or 1 year, then the question of how long will each account survive gets transformed into "within 6 month or 1 year, what is the attrition probability", a problem that logistic regression can handle.

They are often related by the type of problem they are pressed to solve, but the underlying thinking, view of the problem, and mathematics are quite different.