Timeline for Training a Logistic Regression Model
Current License: CC BY-SA 2.5
6 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jul 29, 2010 at 3:47 | comment | added | JoFrhwld | What you want to do is a proportional odds logistic regression. I'm not exactly sure what to call what you have done, | |
| Jul 28, 2010 at 23:06 | vote | accept | Justin Bozonier | ||
| Jul 28, 2010 at 23:06 | comment | added | Justin Bozonier | Thanks guys! TONS O' INFO. :D So then what I'm doing now isn't exactly logistic regression and is instead proportional odds logistic regression? | |
| Jul 28, 2010 at 19:42 | comment | added | JoFrhwld |
Agreed. I just answered in terms of the question, which was about logistic regression. I've fit proportional odds logistic regressions for ordered data like this in R using MASS:polr.
|
|
| Jul 28, 2010 at 19:23 | comment | added | russellpierce | I'd argue from a standpoint of trying to create good guess as to what video game a person likes coding 1s and 0s for like and didn't like isn't a good approach. A scale of 1-5 or 1-7 is quite easy to elicit and will require fewer datapoints to generate a good model (because each data point provides more information). With caveats about treating ordinal data as interval data of course applying, but probably not really that important in this context. | |
| Jul 28, 2010 at 17:01 | history | answered | JoFrhwld | CC BY-SA 2.5 |