Processamento de dados em tempo real: uma análise da relação entre saúde e estilo de vida e o sono

Abstract

The objective of this work is to configure a virtualized environment for real-time data processing, using technologies such as Docker, Hadoop, Apache Kafka, Spark and PostgreSQL. As a demonstration of the viability of the infrastructure, a dataset was chosen to be analyzed and used to train machine learning models. The database presents the relationship between health indicators, lifestyle and sleep disorders. The environment infrastructure was developed with Docker containers, which simplifies the integration of services and tools used. Python was the tool used to perform the analyses that show the existing relationships in the data. The two trained classification algorithms were a Decision Tree (DA) and a neural network Multi layer perceptron (MLP), which presented satisfactory performances. This project exposed how technological tools for real-time data processing can generate flexible, scalable and powerful solutions in a society that daily produces abundant amounts of data.