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
- /
- SAS Programming
- /
- Base SAS Programming
- /
- Linear regression Model 2 dummy variables of same ...

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

03-29-2017 04:37 PM

I am using PROc Reg (Selection = stepwise ) linear regression for banking data which has credit customer information for which we need to predict the factor affecting the usageof credit card. In output of PRoc Reg . My model is giving of the final factors(dummyvariable factors) as

dummy_card1 meaning( amex)

dummy_card4 meaning(discover), (both are types of primary ) refer to screenshot for more clarity

these card are primary credit cards

My question is can we have a final model in which we can have 2 dummy variables from same categorical variables .

Also , can we reduce the number of variable even after the output of ( selection = stepwise ). ?

P.s : I am naive in the world of analytics ..Thanx in advance

Accepted Solutions

Solution

03-30-2017
01:34 PM

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

Posted in reply to pathakvishal

03-30-2017 08:27 AM

pathakvishal wrote:

I am using PROc Reg (Selection = stepwise ) linear regression for banking data which has credit customer information for which we need to predict the factor affecting the usageof credit card. In output of PRoc Reg . My model is giving of the final factors(dummyvariable factors) as

dummy_card1 meaning( amex)

dummy_card4 meaning(discover), (both are types of primary ) refer to screenshot for more clarity

these card are primary credit cards

My question is can we have a final model in which we can have 2 dummy variables from same categorical variables .

Yes, of course you can.

Also , can we reduce the number of variable even after the output of ( selection = stepwise ). ?

Yes, you can do this too, but I wouldn't, because stepwise provides the result that if finds to be the best fit model. I would however recommend that stepwise isn't highly regarded by most statisticians these days, and perhaps something like PROC PLS might be a better alternative with multiple input variables that are correlated.

All Replies

Solution

03-30-2017
01:34 PM

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

Posted in reply to pathakvishal

03-30-2017 08:27 AM

pathakvishal wrote:

I am using PROc Reg (Selection = stepwise ) linear regression for banking data which has credit customer information for which we need to predict the factor affecting the usageof credit card. In output of PRoc Reg . My model is giving of the final factors(dummyvariable factors) as

dummy_card1 meaning( amex)

dummy_card4 meaning(discover), (both are types of primary ) refer to screenshot for more clarity

these card are primary credit cards

My question is can we have a final model in which we can have 2 dummy variables from same categorical variables .

Yes, of course you can.

Also , can we reduce the number of variable even after the output of ( selection = stepwise ). ?

Yes, you can do this too, but I wouldn't, because stepwise provides the result that if finds to be the best fit model. I would however recommend that stepwise isn't highly regarded by most statisticians these days, and perhaps something like PROC PLS might be a better alternative with multiple input variables that are correlated.

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

Posted in reply to PaigeMiller

03-30-2017 01:33 PM

Thanx Mate !!

PaigeMiller wrote:

pathakvishal wrote:

dummy_card1 meaning( amex)

dummy_card4 meaning(discover), (both are types of primary ) refer to screenshot for more clarity

these card are primary credit cards

Yes, of course you can.

Also , can we reduce the number of variable even after the output of ( selection = stepwise ). ?

Yes, you can do this too, but I wouldn't, because stepwise provides the result that if finds to be the best fit model. I would however recommend that stepwise isn't highly regarded by most statisticians these days, and perhaps something like PROC PLS might be a better alternative with multiple input variables that are correlated.