Timeline for Transforming variables for multiple regression in R
Current License: CC BY-SA 3.0
22 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Apr 13, 2017 at 12:44 | history | edited | CommunityBot |
replaced http://stats.stackexchange.com/ with https://stats.stackexchange.com/
|
|
| Sep 18, 2016 at 3:09 | comment | added | StatguyUser | vif() in car package does not seem to work for gam(), so you'll have to check multi-collinearity by writing your own function | |
| Jul 6, 2015 at 20:12 | history | edited | COOLSerdash | CC BY-SA 3.0 |
Fixed hyperlink to Stata-Blog.
|
| Jul 4, 2013 at 17:32 | history | edited | COOLSerdash | CC BY-SA 3.0 |
added 72 characters in body
|
| Jun 9, 2013 at 22:45 | vote | accept | zgall1 | ||
| Jun 9, 2013 at 22:45 | vote | accept | zgall1 | ||
| Jun 9, 2013 at 22:45 | |||||
| Jun 8, 2013 at 19:05 | history | edited | COOLSerdash | CC BY-SA 3.0 |
edited body
|
| Jun 8, 2013 at 19:00 | history | edited | COOLSerdash | CC BY-SA 3.0 |
added 313 characters in body
|
| Jun 8, 2013 at 17:07 | history | edited | COOLSerdash | CC BY-SA 3.0 |
added 534 characters in body
|
| Jun 8, 2013 at 17:02 | comment | added | COOLSerdash |
@zgall1 Thanks for your feedback, I appreciate it. Hm, yes, the transformations didn't seem to have helped much :). At this point, I would probabily try to use splines for the predictors using generalized additive models (GAMs) with the mgcv package and gam. If that doesn't help, I'm at my wit's end I'm afraid. There are people here that are far more experienced than me and maybe they can give you further advice. I am also not knowledgeable with baseball. Maybe there is a more logical model that makes sense with these data.
|
|
| Jun 8, 2013 at 16:55 | history | edited | COOLSerdash | CC BY-SA 3.0 |
added 1260 characters in body
|
| Jun 8, 2013 at 16:54 | comment | added | zgall1 | I wanted to make clear how much I appreciate all of this help. It has been invaluable. | |
| Jun 8, 2013 at 16:54 | comment | added | zgall1 | @COOLSerdash I also tried the GLM method and got the following - i.imgur.com/FjjAdoW.jpg The same comment as the one above applies. There is a clear improvement but I don't fully understand how to proceed from this point. | |
| Jun 8, 2013 at 16:52 | comment | added | zgall1 | @COOLSerdash Using your detailed walkthrough, I applied the Box Cox transformation to my dependent and then independent variables and have the following plot of my diagnostic variables - i.imgur.com/eO01djl.jpg Clearly, there is an improvement but there still seems to be issues with constant variability and unbiasedness and there is definitely an issue with normality. Where can I go from here? | |
| Jun 8, 2013 at 15:58 | history | edited | COOLSerdash | CC BY-SA 3.0 |
added 27 characters in body
|
| Jun 8, 2013 at 15:51 | history | edited | COOLSerdash | CC BY-SA 3.0 |
added 46 characters in body
|
| Jun 8, 2013 at 15:46 | comment | added | zgall1 | Thank you so much for the detailed explanation. I will try and apply it to my data now. | |
| Jun 8, 2013 at 15:46 | history | edited | COOLSerdash | CC BY-SA 3.0 |
added 46 characters in body
|
| Jun 8, 2013 at 15:44 | comment | added | COOLSerdash | @NickCox Thanks (+1 for your answer, btw). The statement that Box-Cox is the most common method comes from John Fox's book. I took it at face value as I don't have enough experience to judge the statement. I'll remove the statement. | |
| Jun 8, 2013 at 15:40 | history | edited | COOLSerdash | CC BY-SA 3.0 |
added 46 characters in body
|
| Jun 8, 2013 at 15:39 | comment | added | Nick Cox | Good explanation. I don't know that explicit Box-Cox is really the most common method of choosing a transformation. If you count people who just choose logs any way, my own wild guess is that it's a minority method. That picky point doesn't affect anything else, naturally. | |
| Jun 8, 2013 at 15:33 | history | answered | COOLSerdash | CC BY-SA 3.0 |
