ggmice (version 0.1.1)

plot_pred: Plot the predictor matrix of an imputation model

Description

Plot the predictor matrix of an imputation model

Usage

plot_pred(
  data,
  vrb = "all",
  method = NULL,
  label = TRUE,
  square = TRUE,
  rotate = FALSE
)

Value

An object of class ggplot2::ggplot.

Arguments

data

A predictor matrix for mice, typically generated with mice::make.predictorMatrix or mice::quickpred, or an object of class mice::mids.

vrb

String, vector, or unquoted expression with variable name(s), default is "all".

method

Character string or vector with imputation methods.

label

Logical indicating whether predictor matrix values should be displayed.

square

Logical indicating whether the plot tiles should be squares.

rotate

Logical indicating whether the variable name labels should be rotated 90 degrees.

Details

The predictor matrix in mice::mice determines the role an imputation model predictor takes in the imputation model. The rows correspond to incomplete target variables, and the columns to imputation model predictors.

A value of 1 indicates that the column variable is a predictor to impute the target (row) variable. The value 0 means that it is not used as predictor.

Imputation methods for multilevel data use other codes than 0 and 1:

• Methods 2l.bin, 2l.lmer, 2l.norm, 2l.pan, 2lonly.mean, 2lonly.norm and 2lonly.pmm use code -2 to indicate the class variable;

• Methods 2l.bin, 2l.lmer, 2l.norm and 2l.pan use code 2 to indicate the random effects;

• Method 2l.pan uses codes 3 and 4 to add class means to codes 1 and 2 respectively.

References

van Buuren, S. (2018). Flexible imputation of missing data. Chapman and Hall/CRC. stefvanbuuren.name/fimd

Examples

# generate a predictor matrix
pred <- mice::quickpred(mice::nhanes)

# plot predictor matrix for all columns
plot_pred(pred)

# plot predictor matrix for specific columns by supplying a character vector
plot_pred(pred, c("chl", "hyp"))

# plot predictor matrix for specific columns by supplying unquoted variable names
plot_pred(pred, c(chl, hyp))

# plot predictor matrix for specific columns by passing an object with variable names
# from the environment, unquoted with `!!`
my_variables <- c("chl", "hyp")
plot_pred(pred, !!my_variables)
# object with variable names must be unquoted with `!!`
try(plot_pred(pred, my_variables))

# plot predictor matrix of mids object
imp <- mice::mice(mice::nhanes, print = FALSE)
plot_pred(imp)