Timeline for Model selection and model performance in logistic regression
Current License: CC BY-SA 3.0
11 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jan 10, 2014 at 22:54 | history | edited | Scortchi♦ | CC BY-SA 3.0 |
fixed typos
|
| Aug 31, 2012 at 8:12 | comment | added | mael | Nonlinear least square. BMA only does logisitc regression if you specify your model using glm (general linear model), from what I can understand at least. But I'm using nls because I'm specifying the model slightly different from how it is in glm. | |
| Aug 30, 2012 at 23:25 | comment | added | probabilityislogic | what is nls? I was pretty sure BMA does logistic regression. | |
| Aug 30, 2012 at 8:50 | comment | added | mael | Thanks, but I'm using nls for my model...so BMA doesn't work. | |
| Aug 29, 2012 at 23:41 | comment | added | probabilityislogic | The BMA package does this for BIC. It is then fairly simple to adjust these probabilities to make it AIC, if that is what you're after. In any event this would not be hard to code up yourself with the glm() function. | |
| Aug 29, 2012 at 12:44 | comment | added | mael | Sorry for my late comment, but do you know of any easy way to calculate this in R? I have the AIC:s in a list or matrix. I'm fairly new to R so any complicated function building is hard. Thanks! | |
| Aug 11, 2012 at 23:28 | comment | added | probabilityislogic | I also agree, I think Bayes Factors are best used for issues of model structure, such as whether to use normal or t distribution for example. They are not useless for covariate selection, but inefficient compared to shrinkage. | |
| Aug 11, 2012 at 13:37 | comment | added | Frank Harrell | Well put. But I think of most model selection exercises as unnecessary (i.e., parsimony is not your friend) and the result of having no priors at all. | |
| Aug 11, 2012 at 2:39 | comment | added | probabilityislogic | Is there a process which doesn't have extreme multiplicity in model selection? You are dealing with a massive discrete space - this invariably leads to a large number of comparisons. I think the question is more whether or not the implicit prior over the models is a reasonable one. | |
| Aug 10, 2012 at 12:53 | comment | added | Frank Harrell | Using AIC on all possible models is a process with extreme multiplicity for which I wonder about the performance. Speaking in broad generalities it is not always logical to think about this as a variable selection problem but rather as a penalization (shrinkage) problem. | |
| Aug 10, 2012 at 8:14 | history | answered | probabilityislogic | CC BY-SA 3.0 |