Linear Regression and Logistic Regression with Missing Covariates

Estimate parameters of linear regression and logistic regression with missing covariates with missing data, perform model selection and prediction, using EM-type algorithms. Jiang W., Josse J., Lavielle M., TraumaBase Group (2020) .

Introduction

misaem is an implementation of methodology which performs statistical inference for logistic regression model with missing data. This method is based on likelihood, including:

1. Estimate the parameters of logistic regression by a stochastic approximation version of EM algorithm;
2. Estimation of parameters' variance based one Louis formula;
3. Model selection procedure based on BIC;
4. Prediction on a test set which may contain missing values.

Installation of package

Now you can install the package misaem from CRAN.

install.packages("misaem")

Using the misaem package

Basicly,

1. miss.saem contains the procedure of estimation for parameters, as well as their variance, and observed likelihood.
2. model_selection aims at selecting a best model according to BIC.
3. pred_saem performs prediction on a test set which may contain missing values.

For more details, You can find the vignette, which illustrate the basic and further usage of misaem package:

library(misaem)
vignette('misaem')

Reference

Stochastic Approximation EM for Logistic regression with missing values (2018, Jiang W., Josse J., Lavielle M., Traumabase group)" arxiv link.