Rank

Returns the rank of reduced-rank regression-type models in the VGAM package.

Usage

Rank(object, ...)
    Rank.rrvglm(object, ...)
    Rank.qrrvglm(object, ...)
    Rank.rrvgam(object, ...)

Arguments

object

Some VGAM object, for example, having class rrvglm-class. The class vglm-class is not included since this is not based on reduced-rank regression.

...

Other possible arguments fed into the function later (used for added flexibility for the future).

Details

Regression models based on reduced-rank regression have a quantity called the rank, which is 1 or 2 or 3 etc. The smaller the value the more dimension reduction, so that there are fewer parameters. This function was not called rank() to avoid conflict with rank.

Value

Returns an integer value, provided the rank of the model makes sense.

Author

T. W. Yee.

Note

This function has not been defined for VGLMs yet. It might refer to the rank of the VL model matrix, but for now this function should not be applied to vglm fits.

Examples

pneumo <- transform(pneumo, let = log(exposure.time),
                            x3  = runif(nrow(pneumo)))
(fit1 <- rrvglm(cbind(normal, mild, severe) ~ let + x3,
                acat, data = pneumo))
#> Call:
#> rrvglm(formula = cbind(normal, mild, severe) ~ let + x3, family = acat, 
#>     data = pneumo)
#> 
#> Coefficients:
#> (Intercept):1 (Intercept):2           let            x3 
#>    -9.2184386    -3.4105224     2.1508694     0.4721274 
#> 
#> Residual deviance: 4.838786 
#> Log-likelihood: -24.99624 
coef(fit1, matrix = TRUE)
#>             loglink(P[Y=2]/P[Y=1]) loglink(P[Y=3]/P[Y=2])
#> (Intercept)             -9.2184386             -3.4105224
#> let                      2.1508694              0.9650085
#> x3                       0.4721274              0.2118246
constraints(fit1)
#> $`(Intercept)`
#>      [,1] [,2]
#> [1,]    1    0
#> [2,]    0    1
#> 
#> $let
#>           [,1]
#> [1,] 1.0000000
#> [2,] 0.4486598
#> 
#> $x3
#>           [,1]
#> [1,] 1.0000000
#> [2,] 0.4486598
#> 
Rank(fit1)
#> [1] 1