- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content
Hello, I am working on a project and have few doubts regarding linear regression,
Q1 How to interpret goodness of fit in proc genmod?
Q2 Which is the best Proc statement for multiple linear regression, which contains categorical variables as covariates?
Accepted Solutions
- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content
@Unnati25 wrote:
Hello, I am working on a project and have few doubts regarding linear regression,
Q1 How to interpret goodness of fit in proc genmod?
This is all explained here: https://documentation.sas.com/?docsetId=statug&docsetTarget=statug_genmod_details01.htm&docsetVersio...
If there is one or two in particular you need assistance with, please provide the specific statistics that you are looking at.
Q2 Which is the best Proc statement for multiple linear regression, which contains categorical variables as covariates?
There is no "BEST". PROC GLM does all of these things, as does PROC MIXED and PROC GLIMMIX, and probably others. It depends on what you want to do. GLM uses ordinary least squares to fit the model, and assumes the errors are independent and normally distributed. MIXED and GLIMMIX use maximum likelihood estimation, and allow for more complex error structures where errors can be correlated.
Paige Miller
- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content
@Unnati25 wrote:
Hello, I am working on a project and have few doubts regarding linear regression,
Q1 How to interpret goodness of fit in proc genmod?
This is all explained here: https://documentation.sas.com/?docsetId=statug&docsetTarget=statug_genmod_details01.htm&docsetVersio...
If there is one or two in particular you need assistance with, please provide the specific statistics that you are looking at.
Q2 Which is the best Proc statement for multiple linear regression, which contains categorical variables as covariates?
There is no "BEST". PROC GLM does all of these things, as does PROC MIXED and PROC GLIMMIX, and probably others. It depends on what you want to do. GLM uses ordinary least squares to fit the model, and assumes the errors are independent and normally distributed. MIXED and GLIMMIX use maximum likelihood estimation, and allow for more complex error structures where errors can be correlated.
Paige Miller
- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content
But we need to get the parameters estimates to write the linear regression equation and then interpret it accordingly. Also can you please explain how to interpret the GLM results.
- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content
@Unnati25 wrote:
Thank you for your response ,
But we need to get the parameters estimates to write the linear regression equation and then interpret it accordingly. Also can you please explain how to interpret the GLM results.
Use the SOLUTION option of the MODEL statement to get the parameter estimates.
You ask a very non-specific question about interpreting the results, so I cannot give a specific explanation. Show us the results, and explain what the problem is that you are having interpreting the results.
Paige Miller