Description
Methods for Missing Data.
Description
Supply functions for the creation and handling of missing data as well as tools to evaluate missing data methods. Nearly all possibilities of generating missing data discussed by Santos et al. (2019) <doi:10.1109/ACCESS.2019.2891360> and some additional are implemented. Functions are supplied to compare parameter estimates and imputed values to true values to evaluate missing data methods. Evaluations of these types are done, for example, by Cetin-Berber et al. (2019) <doi:10.1177/0013164418805532> and Kim et al. (2005) <doi:10.1093/bioinformatics/bth499>.
README.md
missMethods
The goal of missMethods is to make the creation and handling of missing data as well as the evaluation of missing data methods easier.
Installation
You can install the released version of missMethods from CRAN with:
install.packages("missMethods")
And the development version from GitHub with:
# install.packages("devtools")
devtools::install_github("torockel/missMethods")
Usage
missMethods mainly provides three types of functions:
delete_functions for generating missing valuesimpute_functions for imputing missing valuesevaluate_functions for evaluating missing data methods
Run help(package = "missMethods") to see all functions. More details for the delete_ functions are given in a vignette (run vignette("Generating-missing-values")).
Example
This is a very basic workflow to generate missing values, impute the generated missing values and evaluate the imputation result:
library(missMethods)
set.seed(123)
ds_comp <- data.frame(X = rnorm(100), Y = rnorm(100))
ds_mis <- delete_MCAR(ds_comp, 0.3)
ds_imp <- impute_mean(ds_mis)
evaluate_imputed_values(ds_imp, ds_comp, "RMSE")
#> [1] 0.5328238