Visualização de séries temporais multidimensionais: uma análise sobre a qualidade do ar em São CarlosSP

Abstract

Time series data are present in a wide variety of applications, and the explorations, analysis, and understanding of time series dynamics are interesting although it is a quite complex problems. Information visualization techniques have been successfully applied in this context taking advantage of human perceptual abilities when transforming abstract data into visual information. In this master dissertation, we present a new visual analytic methodology for multidimensional time series analysis. The aim of the proposed approach is to allow the exploration of the behavior of particulate matter present in the urban atmosphere of São Carlos-SP, in order to understand the synergy between air quality, climate, and respiratory diseases. Moreover, the features related to particulate matter vary considerably over time, assuming different patterns according to the weather conditions. This dynamic shows the importance of analytical tools that enable the exploration of temporal changes of the data. We developed two interactive visualization tools, mainly in JavaScript programming language and D3 library. The tools are portable and can be used in all the modern web browsers, not requiring any software installation. The data analysis results show a decrease in atmospheric pollutant level over the years. The effectiveness and usefulness of the proposed technologies are demonstrated in case studies which involve real data and it is validated by domain experts.