plot.glmmNPML: Plot Diagnostics for objects of class glmmNPML or glmmGQ In npmlreg: Nonparametric Maximum Likelihood Estimation for Random Effect Models

Description

The functions `alldist` and `allvc` produce objects of type `glmmGQ`, if Gaussian quadrature (Hinde, 1982, `random.distribution="gq"`) was applied for computation, and objects of class `glmmNPML`, if parameter estimation was carried out by nonparametric maximum likelihood (Aitkin, 1996a, `random.distribution="np"`). The functions presented here give some useful diagnostic plotting functionalities to analyze these objects.

Usage

 ```1 2 3 4``` ```## S3 method for class 'glmmNPML' plot(x, plot.opt = 15, noformat=FALSE, ...) ## S3 method for class 'glmmGQ' plot(x, plot.opt = 3, noformat=FALSE, ...) ```

Arguments

 `x` a fitted object of class `glmmNPML` or `glmmGQ`. `plot.opt` an integer with values 0 <= `plot.opt` <=15. `noformat` if `TRUE`, then any formatting of the plots is omitted (useful if the user wants to include the plots into a panel of several other plots, possibly generated by other functions). `...` further arguments which will mostly not have any effect (and are included only to ensure compatibility with the generic `plot()`- function.)

Details

See the help pages to alldist and the vignette (Einbeck & Hinde, 2007). It is sufficient to write `plot` instead of `plot.glmmNPML` or `plot.glmmGQ`, since the generic `plot` function provided in R automatically selects the right model class.

Value

For class `glmmNPML`: Depending on the choice of `plot.opt`, a subset of the following four plots:

 `1` Disparity trend. `2` EM Trajectories. `3` Empirical Bayes Predictions against observed response. `4` Individual posterior probabilities.

The number given in `plot.opt` is transformed into a binary number indicating which plots are to be selected. The first digit (from the right!) refers to plot 1, the second one to plot 2, and so on. For example, `plot.opt=4` gives the binary number 0100 and hence selects just plot 3.

For class `glmmGQ`: Depending on the choice of plot.opt, a subset of plots 1 and 3. Again, the number is transformed into binary coding, yielding only the disparity trend for `plot.opt=1`, only the EBP's for `plot.opt=2`, and both plots for `plot.opt=3`.

Author(s)

Jochen Einbeck and John Hinde (2007)

References

Aitkin, M. (1996a). A general maximum likelihood analysis of overdispersion in generalized linear models. Statistics and Computing 6, 251-262.

Einbeck, J., and Hinde, J.: Nonparametric maximum likelihood estimation for random effect models in R. Vignette to R package npmlreg. Type `vignette("npmlreg-v")` to open it.

Hinde, J. (1982). Compound Poisson regression models. Lecture Notes in Statistics 14, 109-121.

`alldist`, `allvc`
 ```1 2 3 4 5 6 7 8``` ```data(galaxies, package="MASS") gal<-as.data.frame(galaxies) galaxy.np4u <- alldist(galaxies/1000~1,random=~1,k=4,tol=0.5,data=gal,lambda=1) predict(galaxy.np4u, type="response") # EBP on scale of responses plot(galaxy.np4u, plot.opt=4) # plots only EBP vs. response plot(galaxy.np4u, plot.opt=3) # gives same output as given by default when executing alldist plot(galaxy.np4u) # gives all four plots. ```