Sistema para análise e previsão estatística de indicadores ambientais de poluição do ar no contexto de Big Data

Abstract

The constant evolution of computing and computational capabilities of hardware devices allows more data to be stored and used in several different ways. However, despite this diversity, all applications share the same objective: to be able to obtain value from available data, processing it and generating more meaningful new data. This new process is called Big Data. In the atmospherical science field, the Big Data is present to feed different needs. Among these needs there is the need to analyze and forecast air pollution data. In this context, this document presents a Big Data system able to efficiently retrieve data, using it in different statistical analysis and forecast methods. The data was obtained from the United States Environmental Protection Agency website, then it was stored in a NoSQL database, which allows the efficient utilization. The analysis and forecast methods were developed using the programming language R, which facilitates the use of time series techniques and models. The final version of the system has a web interface to access the analyses. At the end, case studies are presented to demonstrate possible analyzes that can be executed in the system. Also, suggestions for future work are presented in order to continue the research that begins in this document.