Optimizing Diabetes Data Insights Through Kmapper-Based Topological Networks: A Decision Analytics Approach for Predictive and Prescriptive Modeling
Year: 2024
Type: article
Abstract: A highly effective technique for identifying and showing structures in high-dimensional datasets is topological data analysis. The Kmapper software creates overlaying clustering graphs and topological networks to facilitate the investigation of such information. The objective of the work was to visualize a dataset of diabetes patients that included information on blood pressure, glucose levels, and pregnancies using the Kmapper software. Afterward, it applied topological data analysis to see if any underlying structures or patterns could be established. The preprocessed dataset of diabetic patients was acquired via Kaggle. Kmapper was run with a difference of parameter settings, which includes 0.4 overlap, 15 hypercubes, and varying numbers of PCA components (1, 2, and 3). We investigated the generated graph visualizations. Although two PCA components were used, the topological graphs disclosed likely intriguing highlights such as three peaks. To understand these illustrated structures in the background of the diabetes data, additional investigation is mandatory. Also with every aspect considered, Kmapper worked well for using topological representations to contribute intuition into the high-dimensional dataset.
Source: Management Science Advances.
DOI: 10.31181/msa1120241
Authors: Muhammad Abid, Muhammad Saqlain
Institutions: North Carolina State University, Northwest University
Language: English
Subfield: Computational Theory and Mathematics
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W4404527261