Subtractive Clustering Method (SCM)

A computationally efficient, density‑based algorithm for estimating cluster centers to initialize fuzzy inference systems.

🔍 Overview

The Subtractive Clustering Method (SCM) algorithm estimates cluster centers by computing each point’s density‑based “potential” within a given radius, then iteratively selecting the highest‑potential points and suppressing their neighbors until a set of well‑spaced centers emerges—without needing to specify the number of clusters in advance.

⚙️ Class Definition

class soft_clustering.NOCD(
    ra: float = 0.5
    ea: float = 0.5,
    er: float = 0.15
)

🔗 Source on GitHub

📋 Parameters

Parameter	Type	Default	Description
`ra`	`float`	`0.5`	Neighborhood radius used to compute each point’s potential.
`ea`	`float`	`0.5`	Acceptance ratio; terminates clustering when potential drops below this.
`er`	`float`	`0.15`	Rejection ratio (reserved for extension; currently unused in core logic).

🚀 Usage Examples

from soft_clustering import SCM
import numpy as np

# Sample 2D data with two clusters
X = np.array([
    [1.0, 2.0], [1.5, 1.8], [1.2, 2.2], [0.9, 1.7],
    [8.0, 8.0], [8.5, 8.2], [9.0, 7.8], [7.5, 8.1]
])

# Initialize and fit the model
model = SCM()
centers = model.fit(X)

print("Cluster centers found by SCM:")
print(centers)

🛠️ Methods

`fit(X)`

Apply subtractive clustering to the input data and return cluster centers.

Parameters:

X (np.ndarray, shape (n_samples, n_features)): Input data points.

Returns:

centers (np.ndarray, shape (n_clusters, n_features)): Coordinates of the discovered cluster centers.

🔗 Source definition

📚 Reference

Chiu, S. L. (1994). Fuzzy Model Identification Based on Cluster Estimation. In Proceedings of the IEEE International Conference on Fuzzy Systems. ACM Digital Library 2656640.