indivplot: plot aggregated COD distribution

Description

Produce a bar plot of the aggregated COD distribution as approximate CSMFs for a fitted "insilico" object.

Usage

indivplot(
  x,
  type = c("errorbar", "bar")[1],
  top = 10,
  causelist = NULL,
  which.plot = NULL,
  xlab = "Causes",
  ylab = "COD distribution",
  title = "COD distributions for the top causes",
  horiz = TRUE,
  angle = 60,
  fill = "lightblue",
  err_width = 0.4,
  err_size = 0.6,
  point_size = 2,
  border = "black",
  bw = FALSE,
  ...
)

Arguments

x: object from get.indiv function.
type: An indicator of the type of chart to plot. "errorbar" for line plots of only the error bars on single population; "bar" for bar chart with error bars on single population.
top: The number of top causes to plot. If multiple sub-populations are to be plotted, it will plot the union of the top causes in all sub-populations.
causelist: The list of causes to plot. It could be a numeric vector indicating the position of the causes in the InterVA cause list (see causetext), or a vector of character string of the cause names. The argument supports partial matching of the cause names. e.g., "HIV/AIDS related death" could be abbreviated into "HIV"; "Other and unspecified infect dis" could be abbreviated into "Other and unspecified infect".
which.plot: Specification of which group to plot if there are multiple.
xlab: Labels for the causes.
ylab: Labels for the CSMF values.
title: Title of the plot.
horiz: Logical indicator indicating if the bars are plotted horizontally.
angle: Angle of rotation for the texts on x axis when horiz is set to FALSE
fill: The color to fill the bars when type is set to "bar".
err_width: Size of the error bars.
err_size: Thickness of the error bar lines.
point_size: Size of the points.
border: The color to color the borders of bars when type is set to "bar".
bw: Logical indicator for setting the theme of the plots to be black and white.
...: Not used.

Author

Zehang Li, Tyler McCormick, Sam Clark

Maintainer: Zehang Li <lizehang@uw.edu>

References

Tyler H. McCormick, Zehang R. Li, Clara Calvert, Amelia C. Crampin, Kathleen Kahn and Samuel J. Clark Probabilistic cause-of-death assignment using verbal autopsies, Journal of the American Statistical Association (2016), 111(515):1036-1049.

Examples

Run this code


if (FALSE) {
# Toy example with 1000 VA deaths
data(RandomVA1)
fit1<- insilico(RandomVA1, subpop = NULL,  
              Nsim = 1000, burnin = 500, thin = 10 , seed = 1,
              auto.length = FALSE)
summary(fit1, id = "d199")

# update credible interval for individual probabilities to 90%
indiv.new <- get.indiv(fit1, CI = 0.9)
fit1$indiv.prob.lower <- indiv.new$lower
fit1$indiv.prob.upper <- indiv.new$upper
fit1$indiv.CI <- 0.9
summary(fit1, id = "d199")


# get empirical aggregated COD distribution 
agg.csmf <- get.indiv(data = RandomVA2, fit1, CI = 0.95, 
                      is.aggregate = TRUE, by = NULL)
head(agg.csmf)

# aggregate individual COD distribution by sex and age
# note the model was fitted assuming the same CSMF for all deaths
# this aggregation provides an approximate CSMF for each sub-groups
agg.by.sex.age <- get.indiv(data = RandomVA2, fit1, CI = 0.95, 
                        is.aggregate = TRUE, by = list("sex", "age"))
head(agg.by.sex.age$mean)

# plot of aggregated individual COD distribution
# 0. plot for all data
indivplot(agg.csmf, top = 10)
# 1. plot for specific one group
indivplot(agg.by.sex.age, which.plot = "Men 60-", top = 10)
# 2. comparing multiple groups
indivplot(agg.by.sex.age, which.plot = list("Men 60+", "Men 60-"), 
                          top = 5)
# 3. comparing multiple groups on selected causes
indivplot(agg.by.sex.age, which.plot = list("Men 60-", "Women 60-"), 
                          top = 0, causelist = c(
                            "HIV/AIDS related death", 
                            "Pulmonary tuberculosis", 
                            "Other and unspecified infect dis", 
                            "Other and unspecified NCD"))
}