Replication Package for Machine Learning Prediction of Parental Stress Through Family Media Ecology

This replication package provides the computational workflow associated with the manuscript Machine Learning Prediction of Parental Stress Through Family Media Ecology. It contains a fully synthetic dataset, documented Jupyter notebooks, a variable dictionary, software requirements, and scripts that generate the machine learning and logistic regression outputs presented in the manuscript and supporting information. The workflow includes repeated stratified train/test evaluation, threshold selection with false-positive control, permutation importance, standardized logistic regression, and bootstrap uncertainty estimation. The synthetic data contain no participant records and are intended solely to demonstrate the analytical procedures and output structure. Numerical results generated from the package will therefore differ from those reported in the manuscript. Access to the original study data is governed by the CAFE Consortium’s data-sharing procedures and requirements.