PRAMS

This repo contains scripts used in the paper Machine Learning-Based Predictive Modeling of Postpartum Depression. We trained machine learning models to predict postpartum depression, and further analysed our models to find features that are important in the development of this phenotype.

software

1. create a conda environment: conda env create -f software/environment.yaml
2. install additional packages: Rscript software/packages.R

scripts and usage

modules: contains helper functions used in this project; if you intend to reproduce these results, you don't need to run this script.

0. helpers.R

preprocess: contains scripts used to preprocess the data

1. preprocessing.R
2. balancing_dataset.R
3. collect_and_split_data.R

train:

4. train_set_one.R
5. train_set_two.R
6. train_set_three.R

analyse:

7. analyse.R

Notes

• the feature_details.csv file was not created by us but it is available