Learning Order Forest for Qualitative-Attribute Data Clustering

Researchers developed a new machine learning method called Learning Order Forest that improves clustering of qualitative data by using tree-like structures to represent relationships between categorical attributes. The joint learning mechanism iteratively optimizes both tree structures and clusters, outperforming 10 competing methods across 12 benchmark datasets.

• →Traditional Euclidean distance clustering fails to capture relationships in qualitative attributes like symptoms or marital status.
• →The new method represents qualitative data relationships using tree-like structures where each value is a vertex.
• →A joint learning mechanism iteratively optimizes both forest structure and clustering results simultaneously.
• →The approach outperformed 10 competing methods across 12 real-world benchmark datasets with statistical significance.
• →The latent distance space of datasets can be effectively represented by a forest of learned trees.