Clusterização e visualização espaço-temporal de dados georreferenciados adaptando o algoritmo marker clusterer: um caso de uso em Curitiba

Abstract

Fifty percent of the world's population live in cities, and the expectation until 2050 is that it reaches 70% (WHO, 2014). Cities consume 75% of the world's natural resources and energy, and generate 80% of greenhouse gases responsible for the greenhouse effect, considering that they occupy only 2% of the world's territory (Signori, 2008). Cities are also the scene of most of the global environmental problems (Gomes, 2009), and it is in the urban context where the social, economic and environmental dimension converge more intensely (European Commission, 2007). This population growth has social, economic and environmental influences that represent a great challenge for the sustainable development of urban planning. The concepts of geographic information systems, smart cities, open data, clustering and data visualization algorithms allow us to understand several questions regarding urban activity in cities, especially, understand the variable "where" things happen. For example: where there is traffic and what time is the most frequent, where it is necessary to perform residential, commercial, industrial standby modeling according to population growth for the land use plan, what are the types of businesses that grew the most in each neighborhood and what is the relationship between them. For this purpose, the following thesis presents a web-mobile system that allows us to understand the spatiotemporal and economic growth of the restaurant licenses of districts Centro, Batel and Tatuquara of Curitiba for the last three decades, performing clustering and visualization of a large amount of open georeferenced data. In terms of achieved results, we can highlight: 1) ability to solve computational problems of overlapping points representing business on a map, 2) ability to understand the economic growth of restaurants licences and what is the relationship between different categories and between districts, 3) execution time less than 3 seconds for 99% of the spatial queries executed, 4) 80.8% of users in evaluation phase consider that the proposed solution allows a better identification and visualization of georeferenced data, and 5) it allows the integration of new sources and types of data.