DPpackage (version 1.1-7.4)

predict.HDPMcdensity: Predictive Information for the Dependent Random Probability Measures.

Description

Plot the probability measures arising from a HDPM of normals model for conditional density estimation. Support provided by the NIH/NCI R01CA75981 grant.

Usage

# S3 method for HDPMcdensity
predict(object,pred,i,r,ask=TRUE,nfigr=2,nfigc=2, ...)

Arguments

object

HDPMcdensity fitted model object.

pred

indicator for the values of the predictors, given by the row pred in xpred, for which the conditional densities must be drawn.

i

study indicator.

r

indicator for including (0) or not (1) the common measure.

ask

logical variable indicating whether the plots must be displayed sequentially or not.

nfigr

number of rows in the figure.

nfigc

number of columns in the figure.

...

further arguments to be passed.

Details

Must run HDPMcdensity first to generate posterior simulations.

References

Mueller, P., Quintana, F. and Rosner, G. (2004). A Method for Combining Inference over Related Nonparametric Bayesian Models. Journal of the Royal Statistical Society, Series B, 66: 735-749.

See Also

HDPMcdensity

Examples

# NOT RUN {
    # Data
      data(calgb)
      attach(calgb)
      y <- cbind(Z1,Z2,Z3,T1,T2,B0,B1)
      x <- cbind(CTX,GM,AMOF)
  
      z <- cbind(y,x)

    #  Data for prediction
      data(calgb.pred)
      xpred <- as.matrix(calgb.pred[,8:10])


    # Prior information
      prior <- list(pe1=0.1,
                    pe0=0.1,
                    ae=1,
                    be=1,
                    a0=rep(1,3),
                    b0=rep(1,3),
                    nu=12,
                    tinv=0.25*var(z),
 		  m0=apply(z,2,mean),
                    S0=var(z),
 		  nub=12,
                    tbinv=var(z))		


    # Initial state
      state <- NULL

    # MCMC parameters

      mcmc <- list(nburn=5000,
                   nsave=5000,
                   nskip=3,
                   ndisplay=100)

    # Fitting the model
      fit1 <- HDPMcdensity(formula=y~x,
                          study=~study,
                          xpred=xpred,
                          prior=prior,
                          mcmc=mcmc,
                          state=state,
                          status=TRUE)

    # Posterior inference
      fit1
      summary(fit1)
       
    # Plot the parameters
    # (to see the plots gradually set ask=TRUE)
      plot(fit1,ask=FALSE)

    # Plot the a specific parameters 
    # (to see the plots gradually set ask=TRUE)
      plot(fit1,ask=FALSE,param="eps",nfigr=1,nfigc=2)

    # Plot the measure for each study 
    # under first values for the predictors, xpred[1,]
      predict(fit1,pred=1,i=1,r=1) # pred1, study 1
      predict(fit1,pred=1,i=2,r=1) # pred1, study 2

    # Plot the measure for each study 
    # under second values for the predictors, xpred[2,]
      predict(fit1,pred=2,i=1,r=1) # pred2, study 1
      predict(fit1,pred=2,i=2,r=1) # pred2, study 2

    # Plot the idiosyncratic measure for each study
    # under first values for the predictors, xpred[1,]
      predict(fit1,pred=1,i=1,r=0) # study 1
      predict(fit1,pred=1,i=2,r=0) # study 2

    # Plot the common measure
    # under first values for the predictors, xpred[1,]
      predict(fit1,pred=1,i=0)
# }