Usage: |
sel = glmselect(code,x,y,opt)
|
Input: |
| code | text string, the short code for the model (e.g.
"bilo" for logit or "noid" for ordinary PLM).
|
| x | n x p matrix, the predictor variables.
|
| y | n x 1 vector, the response variables.
|
| opt | optional, a list with optional input. 'glmselect'
allows the same optional parameters as 'glmest'.
Additionally, the following are supported.
|
| opt.fix | r x 1, r < p, numbers of columns which should be
always included in the model.
|
| opt.shm | integer, if exists and =1, some output is produced
which indicates how the selection is going on.
|
| opt.crit | string, either "aic" or "bic", the selection criterion
to use. If not given "aic" is used.
|
Output: |
| best | p x 5, five best models. |
| bestcrit | list, containing criteria for five best models: |
| bestcrit.aic | 1 x 5, AIC's for five best models. |
| bestcrit.bic | 1 x 5, BIC's for five best models. |
| bestord | p x p or p x (p-r) best models for every order. |
| bestordcrit | list, containing criteria for five best models: |
| bestordcrit.aic | 1 x p or 1 x (p-r), AIC's for five best models
for every order. |
| bestordcrit.bic | 1 x p or 1 x (p-r), BIC's for five best models
for every order. |
| bestfit | GLM fit for best model, contains b, bv and stat
as computed by 'glmest'. |