+ Reply to Thread
Results 1 to 1 of 1

Thread: Simple regression model but needs improvement

  1. #1
    Points: 252, Level: 5
    Level completed: 4%, Points required for next Level: 48

    Posts
    3
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Simple regression model but needs improvement




    I have been working on trying to fit a linear regression line to try to predict college gpa where the the prediction variables are high school GPA(X1) , SAT(X2),ACT (X3), and whether your from a rural/urban area(qualitative variable) (X4). So far I have eliminated Act scores because that showed a weak correlation with college gpa. I've kept the rest of the others in the model.
    Attachment 5144
    This is the full step wise regression and X3 has a large p-value so I removed it from the model.
    Attachment 5142
    Name:  plots.jpg
Views: 126
Size:  87.3 KB

    So I chose Y~X1+X2+X4 and show the plots of residuals of this model. From these plots I see the residuals vs each predictor look good to me (with a few outliers) as that is to be expected in this case. The residuals look randomly distributed around the x=0 line. The qq plot looks decent to me. I've tried adding in interaction terms but I get large p-values due to high multicollinearity possibly, which I think I should not add. I also do not think transformations would be appropriate in this case as the residuals show no signs of a trend. I'm trying to develop the best model I can and what I have above is the best I got so far . Any opinions would be great on what I can try would be great. Thanks.


    Im not exactly sure why i get no postings. Its about the third time already....
    Last edited by raidernation; 03-08-2015 at 03:03 AM.

+ Reply to Thread

           




Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts






Advertise on Talk Stats