Lossless Visualization

Lossless visualization aims to represent high-dimensional data in lower dimensions (2D or 3D) without information loss, enabling intuitive exploration and model building by non-experts. Current research focuses on developing novel coordinate systems, such as General Line Coordinates (including variations like Shifted Paired and Tripled Coordinates), to achieve this lossless projection, often incorporating visual rule discovery algorithms like Sequential Rule Generation. This approach facilitates the creation of explainable machine learning models for diverse data types (numeric, categorical, mixed) by allowing users to directly interact with and interpret data visualizations to build and refine models. The ultimate goal is to empower users to develop and understand machine learning models without requiring specialized data science expertise.