Narayanan, ArvindPham, Hien2023-07-272026-09-292023-07-272026-09-292023-04-192023-07-27http://arks.princeton.edu/ark:/88435/dsp01wp988p09khttps://theses-dissertations.princeton.edu/handle/88435/dsp01wp988p09kThe explosive popularity of machine learning (ML) methods in scientific research comes with many concerns in rigor and reproducibility when applied to. This thesis attempts to analyze the typical sources of concerns and how it manifests in issues with reproducibility, examines discussions on potential solutions in setting standards and in education, and provides a collection of educational materials to be used in conjunction with a set of general reporting standards in ML-based science. This work is aimed at researchers who are practicing ML in various fields of research, but secondarily also hopes to assist students learning ML methodology for scientific research, reviewers of studies that applies ML, and journals that accept studies with ML components.application/pdfenPrinceton University Senior Theses