Topological Machine Learning for Mixed Numeric and Categorical Data

Topological data analysis is a relatively new branch of machine learning that excels in studying high-dimensional data, and theoretically known to be robust against noise. Meanwhile, objects with mixed numeric categorical attributes are ubiquitous real-world applications. However, topological methods usually applied point cloud the best our knowledge there no available framework for classification using methods. In this paper, we propose novel method classification. proposed method, use theory from such as persistent homology, persistence diagrams Wasserstein distance study data. The performance demonstrated by experiments on heart disease dataset. Experimental results show outperforms several state-of-the-art algorithms prediction disease.