What to do when K-means clustering fails : a simple yet principled alternative algorithm
- Submitting institution
-
Aston University
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 21359158
- Type
- D - Journal article
- DOI
-
10.1371/journal.pone.0162259
- Title of journal
- PLoS ONE
- Article number
- e0162259
- First page
- -
- Volume
- 11
- Issue
- 9
- ISSN
- 1932-6203
- Open access status
- Deposit exception
- Month of publication
- September
- Year of publication
- 2016
- URL
-
-
- Supplementary information
-
-
- Request cross-referral to
- -
- Output has been delayed by COVID-19
- No
- COVID-19 affected output statement
- -
- Forensic science
- No
- Criminology
- No
- Interdisciplinary
- Yes
- Number of additional authors
-
3
- Research group(s)
-
A - Aston Institute of Urban Technology and the Environment (ASTUTE)
- Citation count
- 36
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- We explain the challenges of one of the most used clustering algorithms (K-means) for a wider audience and motivate a more flexible model-based alternative, which opens many avenues of applications for Bayesian parametric and nonparametric algorithms to resource-constrained problems. The proposed MAP-DP and related extensions have been since adopted from across a wide domain of applications, from Parkinson’s disease phenotyping at the Oxford Parkinson’s Disease Centre (contact: https://www.ndcn.ox.ac.uk/team/michele-hu), to clustering of single-cell data at the University of Leeds and robotics applications at MIT. The mathematics behind this approach was explained in a contemporaneous paper (doi: 10.1214/16-EJS1196), which has been well-cited.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -
