A loose-pattern process approach to clustering fuzzy data sets

Area of Science:

• Data Science
• Computer Science
• Machine Learning

Background:

• Clustering algorithms are essential for data analysis.
• Existing methods may struggle with complex or loosely defined patterns.
• A need exists for flexible clustering approaches.

Purpose of the Study:

• To present a new loose-pattern process approach for clustering sets.
• To introduce two novel tight-pattern clustering methods: GLC and OUPIC.
• To evaluate the effectiveness of the proposed loose-pattern assigning classes method.

Main Methods:

• Loose-pattern rejection based on q-nearest neighbors.
• Tight-pattern classification using GLC and OUPIC methods.
• Loose-pattern class assignment guided by a heuristic membership function.

Main Results:

• The loose-pattern rejection effectively filters data points.
• GLC and OUPIC demonstrated robust tight-pattern classification.
• The heuristic membership function showed promise in assigning classes.

Conclusions:

• The proposed loose-pattern process offers a flexible framework for set clustering.
• The integration of nearest neighbor rules and heuristic functions enhances clustering accuracy.
• Further research can explore broader applications and optimizations.