Authors: Julie Josse (CMAP, Inria), Nicolas Prost (CMAP, Inria), Erwan Scornet (CMAP), Gaël Varoquaux (Inria).

Update: the directory Notebook contains a tutorial on key results of the paper.

Access the Notebok via Binder at the following URL

https://mybinder.org/v2/gh/dirty-data/supervised_missing/master?filepath=Notebook%2FA%20toy%20regression%20model%20with%20missing%20values.ipynb

Notes Without Binder, install the required Python libraries via

pip install -r requirements.txt

This repository contains the code for the paper:

Julie Josse, Nicolas Prost, Erwan Scornet, Gaël Varoquaux. On the consistency of supervised learning with missing values. 2019. 〈hal-02024202〉https://arxiv.org/abs/1902.06931

The directory analysis contains the code for figures 1 and 2 (section 5).

boxplots corresponds to figures 3 and 4 (section 6). There are three separate files: one containing the functions, one containing the script for computation, and two for the visualisation (one of each of the two boxplots).

consistency is used for figure 5 (section 6). There are three files as for the boxplot, but in addition, approximate Bayes rates are computed in bayesrates.R with oracle multiple imputation, as detailed in the paper.

The scripts require the following R packages:

rpart
party
ranger
xgboost
MASS
norm
doParallel
doSNOW
gridExtra
viridis

To run script_boxplots.R or script_consistency.R with, say, 20 jobs to parallelize the "for" loop and 10 threads per forest/boosting, do

Rscript boxplots/script_boxplots.R 20 10
Rscript consistency/script_consistency.R 20 10

To build the figures, just run the scripts,

Rscript boxplots/visualisation_boxplot1.R
Rscript boxplots/visualisation_boxplot2.R
python consistency/visualisation_consistency.py

All figure outputs go to the directory figures (created when nonexistent).

Nicolas Prost

July 10, 2019