The p-median model as a tool for clustering psychological data

Area of Science:

• Data Science
• Computer Science
• Statistics

Background:

• The p-median clustering model offers a combinatorial approach for partitioning data into distinct, non-hierarchical groups.
• Current statistical software packages for social and behavioral sciences lack effective implementations of this model.

Purpose of the Study:

• To introduce and describe the p-median clustering model.
• To discuss available software and their capabilities for implementing p-median clustering.
• To demonstrate the application of p-median clustering to a complex dataset.

Main Methods:

• The p-median clustering model constructs object classes around exemplars (manifest data objects).
• Remaining data instances are assigned to their nearest cluster centers.
• The study details the mechanics of the p-median clustering algorithm.

Main Results:

• The paper presents a comprehensive overview of p-median clustering mechanics.
• It evaluates the capabilities of currently available software for implementing the model.
• The application to a food perception dataset illustrates the model's utility.

Conclusions:

• P-median clustering is a valuable technique for data partitioning.
• Accessible software implementations are crucial for wider adoption in social and behavioral sciences.
• The model effectively handles complex, structured datasets.