A new nonlinear dot plots visualization based on an undirected reassignment algorithm

• Published in: Journal of visualization 2021-04, Vol.24 (2), p.289-300
• Main Authors: Xiao, Yun, Wang, Changqing, Li, Kang, Liu, Baoying, Guo, Jun, Wang, Wei
• Format: Article
• Language: English
• Subjects: Algorithms; Classical and Continuum Physics; Computer Imaging; Data analysis; Density; Engineering; Engineering Fluid Dynamics; Engineering Thermodynamics; Heat and Mass Transfer; Outliers (statistics); Pattern Recognition and Graphics; Regular Paper; Scientific visualization; Vision; Visualization
• ISSN: 1343-8875; 1875-8975
• DOI: 10.1007/s12650-020-00711-5

The original nonlinear dot plots were designed for dots of varying sizes by the two-way sweep algorithm. However, this algorithm uses the average of the starting point of the two-way sweep to determine the column position, which can lead to inaccurate position of the obtained data columns. Meanwhile, the two-way sweep algorithm has the shortcoming of unclear expression in high-density data area. In order to address this problem, we propose an improved nonlinear dot plot based on an undirected reassignment algorithm. The improved nonlinear dot plots can process and assign data in order to achieve an optimized layout that can identify the suitable dot position and data distribution. The proposed algorithm is more effective at displaying outliers and avoiding overlaps. Dots with large differences can be presented in the nonlinear dot plots. Our proposed method can further be combined with other data visualization attributions for data analysis purposes. Using a series of real datasets, the improved nonlinear dot plots are compared with both the conventional dot plots and the original nonlinear dot plots. Results show that our improved nonlinear dot plots not only allow for dots of varying sizes, but also clearly display the data with extremely high density. Graphic abstract