Full text
| |
| Author(s): |
Wilson Estecio Marcilio Junior
Total Authors: 1
|
| Document type: | Master's Dissertation |
| Press: | São José do Rio Preto. 2018-12-20. |
| Institution: | Universidade Estadual Paulista (Unesp). Instituto de Biociências Letras e Ciências Exatas. São José do Rio Preto |
| Defense date: | 2018-12-07 |
| Advisor: | Danilo Medeiros Eler |
| Abstract | |
Multidimensional projections are an important tool for analyzing multidimensional datasets. However, although the graphical representation of multidimensional projection brings benefits according to cluster identification and similarity analysis, such representation presents issues when the number of instances or the dimensionality of the dataset increases. In this work, a multilevel exploration approach in visualizations generated to encode multidimensional projections is presented, in which the goal is to provide subsidies for an exploration with lower cognitive load than the common approaches. The proposed technique is based on selecting representative to provide a context to guide the user in the exploration process, besides using Voronoi diagrams to define clusters. In the experiments, the best suited algorithms to select representative are presented, as well as the impact of different multidimensional projection techniques and the feature space of the analyzed dataset. Finally, two case studies are presented to show how the exploration approach works. (AU) | |
| FAPESP's process: | 16/11707-6 - Employing overlapping removal and representatives identification techniques in visualizations based on multidimensional projection |
| Grantee: | Wilson Estécio Marcílio Júnior |
| Support Opportunities: | Scholarships in Brazil - Master |