Bi-Objective k-Nearest Neighbors Imputation for Multilevel Data

The bi-objective k-nearest neighbors method (biokNN) is an imputation method designed to estimate missing values on data with a multilevel structure. The original algorithm is an extension of the k-nearest neighbors method proposed by Bertsimas et al. (2017) (< https://jmlr.org/papers/v18/17-073.html>) using a bi-objective approach. A brief description of the method can be found in Cubillos (2021) (< https://pure.au.dk/portal/files/214627979/biokNN.pdf>). The 'biokNN' package provides an R implementation of the method for datasets with continuous variables (e.g. employee productivity, student grades) and a categorical class variable (e.g. department, school). Given an incomplete dataset with such structure, this package produces complete datasets using both single and multiple imputation, including visualization tools to better understand the pattern of the missing values.


News

Reference manual

It appears you don't have a PDF plugin for this browser. You can click here to download the reference manual.

install.packages("biokNN")

0.1.0 by Maximiliano Cubillos, 3 months ago


https://github.com/mcubillos3/biokNN


Report a bug at https://github.com/mcubillos3/biokNN/issues


Browse source code at https://github.com/cran/biokNN


Authors: Maximiliano Cubillos [aut, cre]


Documentation:   PDF Manual  


GPL (>= 2) license


Imports dplyr, cluster, mice, stats, magrittr, ggplot2, tidyr, desc, lme4, mitml

Suggests knitr, rmarkdown, testthat


See at CRAN