plot.imputed: Plot proposed imputations and their conditional censored...

In TiStat/Imputegamlss: Imputing censored covariates with distributional regression - gamlss

Description

Given an imputex object, this funciton plots:
(1) The approximate averaged quantiles of the valid part of the censored conditonal distributions, from which the proposed vectors were 'drawn' before the proposals were aggregated via median to become the imputed vector.
(2) The actual imputations for each observation. The red dots are the median proposals, which are in fact the final imputed vector.
(3) Density plots on each proposal vector. Consider that each proposal vector is drawn from a bootmodel-distribution, whose observations have their own parameters. So each proposal vector is a realization of a mixed distribution. Suppose, that the booted models are relatively similar, the density estimates on all proposal vectors (which are displayed) should be relatively similar with little variation.

Usage

## S3 method for class 'imputed'
plot(x, boxes = FALSE, ...)

Arguments

x	Object of class "imputex.
boxes	boolean. Indicating whether (2) should be displayed as a boxplot. Note that the median values are the imputation values.
...	plot-specific arguments. See plot() documentation.

Value

Return figures with information on object of class "imputed".

Examples

# Simulating a dataset
rinterval = simulateData(n= 300,
                         param.formula = list(mu = ~exp(x1), sigma = ~sin(x2)),
                         name = 'x1', subset = ~ x1 > 0.6, prob = 0.8 ,
                         damage =list(c(0.8, 0.99), c(1.2,1.5)),
                         family = 'NO',
                         correlation = NULL)
                         
d  <- imputex(data = rinterval$defected,
              xmu_formula= x1~y,
              indicator = "indicator",
              censtype = 'interval',
              intervalstart = 'lower')
              
plot(d, boxes = FALSE)

TiStat/Imputegamlss documentation built on May 20, 2019, 9:25 a.m.