Developer Friendly and Computationally Efficient Predictive Modeling without Information Leakage: The emil Package for R

Christofer L Bäcklin, Mats G Gustafsson.

This repository contains all code required to run the examples, benchmarking test, and producing the figures of the original publication.

Code examples from the paper

The code examples of section 2 and 3 of the paper can be found in the script code.R.

The file should be run using the makefile command below:

make code-examples

Alternatively, to skip the most time consuming steps you may also run the like this:

R --vanilla -f code.R --args --quick

Benchmark study

Clone, install dependencies, and run in one go:

git clone https://github.com/Molmed/Backlin-2017.git
cd Backlin-2017
make benchmark

The code is written for MacOS or Linux operating systems with R version 3.5.0 and package dependencies listed in dependencies.csv.

In addition to R dependencies you also need Python (preferably version 3.5.2) to enable memory profiling, implemented using syrupy.

The code should work fine with other versions of R, Python, and package dependencies too but is not guaranteed to do so. If you encounter errors please post an issue or even better a pull-request with a fix.