Multi-Dimensional Radial Visualization: A Novel Framework for Interactive Exploration of High-Dimensional Tabular Data

This paper introduces Multi-Dimensional Radial Visualization (MDRV) — a novel, interactive 3D visualization framework designed to explore high-dimensional tabular data (5–7 features) while preserving feature interpretability. Unlike traditional dimensionality reduction techniques (e.g., PCA, t-SNE), MDRV maintains direct mappings of original features using a radial axis layout in three-dimensional space. Each instance is visualized as a radial star projected along the vertical axis, allowing users to examine feature contributions and target variable relationships in an intuitive spatial layout. We evaluate MDRV across four real-world datasets (medical diagnosis, Titanic survival, house price prediction, and delivery time forecasting) and demonstrate its advantages in feature analysis, class separation, and outlier detection. The system supports real-time 3D interaction including rotation, filtering, and axis selection, making it suitable for both expert analysts and general users. The code is open-source and available on GitHub.