Unsupervised Machine Learning for Scientific Discovery: Workflow and Best Practices
Not on the map. Why?
Abstract (arXiv)
Unsupervised machine learning is widely used to mine large, unlabeled datasets to make data-driven discoveries in critical domains such as climate science, biomedicine, astronomy, chemistry, and more. However, despite its widespread utilization, there is a lack of standardization in unsupervised learning workflows for making reliable and reproducible scientific discoveries. In this paper, we present a structured workflow for using unsupervised learning techniques in science. We highlight and discuss best practices starting with formulating validatable scientific questions, conducting robust data preparation and exploration, using a range of modeling techniques, performing rigorous validation by evaluating the stability and generalizability of unsupervised learning conclusions, and promoting effective communication and documentation of results to ensure reproducible scientific discoveries. To illustrate our proposed workflow, we present a case study from astronomy, seeking to refine globular clusters of Milky Way stars based upon their chemical composition. Our case study highlights the importance of validation and illustrates how the benefits of a carefully-designed workflow for unsupervised learning can advance scientific discovery.
Key facts. Cited by 2 papers in the map (data through 7 Oct 2026). Builds on 33 of its 228 references that are in the map. First posted on arXiv 5 Jun 2025.
Builds on
33 of 228 references.
References · 33
- 2018UMAP: Uniform Manifold Approximation and Projection for Dimension Reductioncited by 2,841
- 2018World Modelscited by 1,444
- 2015Causal inference using invariant prediction: identification and confidence intervalscited by 533
- 2019Explainable Machine Learning for Scientific Insights and Discoveriescited by 165
- 2013Stabilitycited by 119
- 2020Hyper-Parameter Optimization: A Review of Algorithms and Applicationscited by 114
- 2020Parametric UMAP embeddings for representation and semi-supervised learningcited by 103
- 2018Tunability: Importance of Hyperparameters of Machine Learning Algorithmscited by 98
- 2021Learning Dynamic Graph Representation of Brain Connectome with Spatio-Temporal Attentioncited by 86
- 2022Exploring the Whole Rashomon Set of Sparse Decision Treescited by 80
- 2019Veridical Data Sciencecited by 65
- 2024Amazing Things Come From Having Many Good Modelscited by 52
21 more
- 2014A survey of dimensionality reduction techniquescited by 50
- 2022Rashomon Capacity: A Metric for Predictive Multiplicity in Classificationcited by 46
- 2022Predictive Multiplicity in Probabilistic Classificationcited by 44
- 2020Selective Inference for Hierarchical Clusteringcited by 37
- 2023Exploring and Interacting with the Set of Good Sparse Generalized Additive Modelscited by 30
- 2020Importance of Tuning Hyperparameters of Machine Learning Algorithmscited by 19
- 2021The Art and Practice of Data Science Pipelines: A Comprehensive Study of Data Science Pipelines In Theory, In-The-Small, and In-The-Largecited by 17
- 2018A Visual Interaction Framework for Dimensionality Reduction Based Data Explorationcited by 15
- 2020Machine Learning Pipelines: Provenance, Reproducibility and FAIR Data Principlescited by 14
- 2024A Comprehensive Survey of Reinforcement Learning: From Algorithms to Practical Challengescited by 13
- 2023Meta-survey on outlier and anomaly detectioncited by 12
- 2018Performance evaluation and hyperparameter tuning of statistical and machine-learning models using spatial datacited by 11
- 2019Goals, Process, and Challenges of Exploratory Data Analysis: An Interview Studycited by 11
- 2019A Systematic Review of Unsupervised Learning Techniques for Software Defect Predictioncited by 4
- 2024Assessing and improving reliability of neighbor embedding methods: a map-continuity perspectivecited by 4
- 2025Are machine learning interpretations reliable? A stability study on global interpretationscited by 3
- 2018Machine learning in APOGEE: Unsupervised spectral classification with K-meanscited by 2
- 2019Deriving star cluster parameters with convolutional neural networks. II. Extinction and cluster/background classificationcited by 2
- 2015The topology of large Open Connectome networks for the human braincited by 1
- 2024Towards Unsupervised Validation of Anomaly-Detection Modelscited by 1
- 2024Rashomon effect in Educational Research: Why More is Better Than One for Measuring the Importance of the Variables?cited by 1
Built on by
Citing papers per year: 2025: 1; 2026: 1.
2 papers.