Same Stats, Different Graphs Made Simple and Fast by Linear Transformations
S. van Wageningen, A.C. Telea, T. Mchedlidze · Utrecht University Repository (Utrecht University) · 2026
While summary statistics are used consistently to report data results, their limitations (and the importance of visualizing data) are well known. Yet, current methods that aim to approximate shapes to match some desired summary statistics are quite complex and can fail the approximation for complex shapes. We present a simple linear transformation LT that performs the same approximation more effectively and efficiently. We show that this method works well for complicated shapes and large datasets, visually outperforms almost all baseline algorithms for most shapes, while being significantly faster and simpler. We also present an iterative approach that improves the method LT in case of extreme statistical values.