https://github.com/gagolews/bfpm_critique
A Critique of the Bounded Fuzzy Possibilistic Method - Supplementary Files
Science Score: 26.0%
This score indicates how likely this project is to be science-related based on various indicators:
-
○CITATION.cff file
-
✓codemeta.json file
Found codemeta.json file -
○.zenodo.json file
-
✓DOI references
Found 16 DOI reference(s) in README -
○Academic publication links
-
○Academic email domains
-
○Institutional organization owner
-
○JOSS paper metadata
-
○Scientific vocabulary similarity
Low similarity (6.1%) to scientific vocabulary
Repository
A Critique of the Bounded Fuzzy Possibilistic Method - Supplementary Files
Basic Info
Statistics
- Stars: 0
- Watchers: 2
- Forks: 0
- Open Issues: 0
- Releases: 0
Metadata Files
README.md
A Critique of the Bounded Fuzzy Possibilistic Method — Supplementary Files
The Bounded Fuzzy Possibilistic Method (BFPM) was introduced in doi:10.1016/j.fss.2019.07.011. I demonstrate that there are some critical flaws in the proposed algorithm, which makes the results presented therein highly questionable. In particular, the method may generate meaningless cluster membership degrees or even fail to converge when run on some classical benchmark data sets.
bfpm.py implements both the BFPM as well as the (now-classic) Fuzzy (weighted) c-means method (FCM). The implementation of the BFPM is straightforward, because it is an arbitrary (faulty) modification of the FCM.
iris.ipynb runs the BFPM on the famous Iris data set. The algorithm does not converge.
yeast.ipynb studies the behaviour of the BFPM on the Yeast data set from the UCI Machine Learning repository. The algorithm converges to a solution representing one trivial cluster (all cluster centres coincide and are equal to the centroid of the whole data set).
unbalance.ipynb provides a simple illustration
that a solution identified by the FCM does not necessarily correspond
to the global minium of the underlying objective function.
Even multiple restarts from different random initial guesses may not
be enough, although they can improve the solution significantly.
This is a well known behaviour that we also observe in the case
of the classic k-means method (compare, e.g., the nstart
argument to the kmeans() function in R or the
n_init argument in Python's sklearn.cluster.KMeans).
References
H. Yazdani, Bounded fuzzy possibilistic method, Fuzzy Sets and Systems 389 (2020) 51–65. doi:10.1016/j.fss.2019.07.011
J. Bezdek, R. Ehrlich, W. Full, FCM: The fuzzy c-means clustering algorithm, Computers & Geosciences 10 (1984) 191–203. doi:10.1016/0098-3004(84)90020-7
M. Gagolewski, A critique of the Bounded Fuzzy Possibilistic Method, Fuzzy Sets and Systems 426 (2022) 176-181. doi:10.1016/j.fss.2021.07.001
Owner
- Name: Marek Gagolewski
- Login: gagolews
- Kind: user
- Location: Melbourne, VIC, Australia
- Company: Deakin University
- Website: https://www.gagolewski.com
- Repositories: 23
- Profile: https://github.com/gagolews
Free universities!
GitHub Events
Total
Last Year
Issues and Pull Requests
Last synced: 9 months ago
All Time
- Total issues: 0
- Total pull requests: 0
- Average time to close issues: N/A
- Average time to close pull requests: N/A
- Total issue authors: 0
- Total pull request authors: 0
- Average comments per issue: 0
- Average comments per pull request: 0
- Merged pull requests: 0
- Bot issues: 0
- Bot pull requests: 0
Past Year
- Issues: 0
- Pull requests: 0
- Average time to close issues: N/A
- Average time to close pull requests: N/A
- Issue authors: 0
- Pull request authors: 0
- Average comments per issue: 0
- Average comments per pull request: 0
- Merged pull requests: 0
- Bot issues: 0
- Bot pull requests: 0