Fuzzy C-Means Clustering for Ordinal Data using Triangular Fuzzy Numbers
The fcmfd package implements fuzzy clustering for ordinal Likert-type data using Triangular Fuzzy Numbers (TFNs).
It is designed for datasets where responses are measured on discrete ordinal scales (e.g., 1-5, 1–7, 1-10 or 0–10), providing a robust alternative to traditional clustering approaches.
- Fuzzy C-Means clustering adapted to TFNs
- Automatic selection of the optimal number of clusters
- Xie–Beni validity index
- Support for Likert-type data
- Cluster assignment and prototype extraction
- Visualization tools
# install.packages("devtools")
devtools::install_github("yourusername/fcmfd")library(fcmfd)
# Load dataset
data(sim_likert_0_10)
# Run clustering
result <- fcmTFN(
data = sim_likert_0_10,
option = "B",
k_values = 2:6
)
# Summary
summary(result)
# Cluster assignment
clusters <- cluster_assignment(result)
table(clusters)
# Plot Xie–Beni index
plot_xb(result)-
sim_likert7 Simulated dataset with a 1–7 Likert scale
-
sim_likert_0_10 Simulated dataset with a 0–10 Likert scale and latent cluster structure
The package combines:
- Fuzzy C-Means clustering
- Triangular Fuzzy Numbers representation
- Xie–Beni cluster validity index
to provide a framework tailored for ordinal data.
- Survey analysis
- Social sciences
- Customer satisfaction
- Quality of life studies
- Likert-type data clustering
Coppi, R., D’Urso, P., & Giordani, P. (2011). Fuzzy clustering of fuzzy data. Computational Statistics & Data Analysis. https://doi.org/10.1016/j.csda.2010.09.013
Xie, X. L., & Beni, G. (1991). A validity measure for fuzzy clustering. IEEE Transactions on Pattern Analysis and Machine Intelligence. https://doi.org/10.1109/34.85677
José Ortigas
MIT