Uncovering Representative Groups in Multidimensional Projections

Abstract

Multidimensional projection-based visualization methods typically rely on clustering and attribute selection mechanisms to enable visual analysis of multidimensional data. Clustering is often employed to group similar instances according to their distance in the visual space. However, considering only distances in the visual space may be misleading due to projection errors as well as the lack of guarantees to ensure that distinct clusters contain instances with different content. Identifying clusters made up of a few elements is also an issue for most clustering methods. In this work we propose a novel multidimensional projection-based visualization technique that relies on representative instances to define clusters in the visual space. Representative instances are selected by a deterministic sampling scheme derived from matrix decomposition, which is sensitive to the variability of data while still been able to handle classes with a small number of instances. Moreover, the sampling mechanism can easily be adapted to select relevant attributes from each cluster. Therefore, our methodology unifies sampling, clustering, and feature selection in a simple framework. A comprehensive set of experiments validate our methodology, showing it outperforms most existing sampling and feature selection techniques. A case study shows the effectiveness of the proposed methodology as a visual data analysis tool.

Supporting Analysis of Dimensionality Reduction Results with Contrastive Learning

“…Among the mentioned studies, the works by Joia et al [36] and Turkay et al [69] are most related to ours in terms of identifying the representative features for each cluster. To identify such features, both methods refer to each cluster's principal components (PCs) computed by PCA (and the correlation between the features and PCs).…”

“…This linking of the DR result and the salient features helps the user interpret the DR result. Similarly, Joia et al [36] linked the DR result and the information of features into one plot. In addition to an automatic selection of clusters, they obtained representative features for each cluster by using PCA.…”

Visual Interaction with Dimensionality Reduction: A Structured Literature Analysis

“…(We intended to exclude theory and evaluation papers, as well as papers focused on unrelated or tangential topics such as volume rendering or physical flow data). Second, we checked if the paper addresses the combination of visualization, DR and interaction, and if the interaction feeds back to the DR. For example Joia et al [32] present an interesting technique for sampling and feature selection. However, there is no interaction that causes a recalculation of the DR.…”

Técnicas de Projeção para Identificação de Grupos e Comparação de Dados Multidimensionais Usando Diferentes Medidas de Similaridade

“….2. Método para Identificação de Grupos com Base em Projeção Column Selection Method (CSM)[Joia et al 2015], um método de visualização apoiado em projeção multidimensional que permite agrupar dados. CSM opera no espaço visual, garantindo que os grupos obtidos não fiquem fragmentados durante a visualização.…”