plot.gerbil {gerbil}

Plotting for gerbil objects

Description

Using a gerbil object as an input, this function gives diagnostic plots for selected variables

Usage

## S3 method for class 'gerbil'
plot(
  x,
  y = NULL,
  type = "Univariate",
  imp = 1,
  col = NULL,
  lty = NULL,
  lwd = NULL,
  pch = NULL,
  log = NULL,
  legend = NULL,
  legend.loc = "topright",
  mfrow = c(3, 2),
  trace.type = "Mean",
  file = NULL,
  sep = FALSE,
  height = NULL,
  width = NULL,
  partial = "imputed",
  ...
)

Arguments

x

A gerbil object containing the imputed data.

y

A vector listing the column names of the imputed data for which plots should be created. See details. By default, y contains all columns of the data that required imputation.

type

A scalar used to specify the type of plots that will be created. Options include univariate (marginal) plots (type = 1), bivariate plots (type = 2), and trace plots (type = 3). See details. Defaults to type = 1.

imp

A scalar or vector indicating which of the multiply imputed datasets should be used for plotting. Defaults to imp = 1. Setting imp = TRUE will include all imputed datasets.

col

The color used for plotting – should be a vector of length equal to imp + 1. The first element references plotting of observed data, and remaining elements reference plotting of imputed data.

lty

The line type used for plotting imputed values with trace lines or density plots – should be a vector of length equal to imp + 1. The first element references plotting of observed data, and remaining elements reference plotting of imputed data.

lwd

The line width used for density and trace line plotting – should be a vector of length equal to imp + 1. The first element references plotting of observed data, and remaining elements reference plotting of imputed data.

pch

A length-2 vector that indicates the plotting symbol to be used for imputed and observed values in scatter and lattice plots.

log

A character vector that includes names of variables of which a log transformation is to be taken prior to plotting.

legend

A character or expression vector to appear in the legend. If FALSE or 'n', no legend is created. Defaults to c("Observed", "Imputed", ...).

legend.loc

The location of the legend in the plots.

mfrow

The layout of plots across a single page when there are to be multiple plots per page (as is the case when file is non-NULL and sep = FALSE).

trace.type

The type of trace plot to be created (only valid when type = 3). See details. Defaults to trace.type = 1.

file

A character string giving the name of file that will be created in the home directory containing plots. The name should have a .pdf or .png extension. If NULL (the default), no file is created.

sep

If sep = TRUE, separate plots will be generated for each outcome. Applicable only if plots are saved to file (plot.file is non-NULL). To change display of plots produced as output, use par.

height

The height of the graphics region (in inches) when a pdf is created.

width

The width of the graphics region (in inches) when a pdf is created.

partial

Indicates how partially imputed pairs are handled in bivariate plotting. If 'imputed', cases with at least one missing variable in a pair are considered imputed. Otherwise (partial = 'observed'), only cases with both variables in the pair missing are considered imputed.

...

Arguments to be passed to methods, such as plot.

Details

Three types of plots may be produced: 1) Univariate (produced by setting type = 1): Compares the marginal distribution of observed and imputed values of a given variable. Density plots are produced for continuous variables, and bar plots are given for binary, categorical, and ordinal variables. For semi-continuous variables, two plots are constructed: a) a bar plot for the binary portion of the variable and 2) a density plot for the continuous portion. 2) Bivariate (produced by setting type = 2): Compares the bivariate distributions of observed and imputed values of two variables. Scatter plots are produced if both variables are continuous or semi-continuous, box plots are produced if one variable is continuous or semi-continuous and the other is not, and a lattice plot is produced if neither variable is continuous or semi-continuous. For bivariate plots, imputed observations are those that have one or more of the values of the pair missing within the original dataset. 3) Trace lines (produced by setting type = 3): Plots a pre-specified parameter across iterations of MCMC in order to examine convergence for a given variable. Parameters that may be plotted include means (trace.type = 1) and variances (trace.type = 2).

Multiple plots may be created, as determined by the variable names listed in the parameter y. For univariate and trace plots, one plot is created for each variable listed in y. For bivariate plotting, one plot is created for each combination of two elements within the vector y (as such, y must have a length of at least two in this case). For trace plotting, elements of y should correspond to column names in the dataset that has been expanded to include binary indicators for categorical and semi-continuous variables. If multiple plots are to be created, it is recommended to specify a file for output using the parameter file, in which case separate files will be created for each plot (if sep = TRUE) or all plots will be written to the same file (if sep = FALSE).

The only required input is a parameter x which is a gerbil object.

Value

No returned value, but instead plots are generated in the workspace or written to a specified directory.

Examples

#Load the India Human Development Survey-II dataset
data(ihd_mcar) 

# Create a gerbil object

imps.gerbil <- gerbil(ihd_mcar, m = 1, ords = "education_level", semi = "farm_labour_days", 
       bincat = "job_field")

# Univariate plotting of all variables to a file
plot(imps.gerbil, type = 1, file = file.path(tempdir(), "gerbil_univariate.pdf"))

# Bivariate plotting of all variables to a file
plot(imps.gerbil, type = 2, file = file.path(tempdir(), "gerbil_bivariate.pdf"))

# Trace plotting of all variables to a file
plot(imps.gerbil, type = 3, file = file.path(tempdir(), "gerbil_ts.pdf"))

# Univariate plotting of one variable (not to a file)
plot(imps.gerbil, type = 1, y = "job_field")

# Bivariate plotting of one pair of variables (not to a file)
plot(imps.gerbil, type = 2, y = c("job_field", "income"))

# Bivariate plotting of one pair of variables (not to a file) with income logged
plot(imps.gerbil, type = 2, y = c("job_field", "income"), log = "income")