ETL and analysis of IoT data using OpenTSDB, Kafka, and Spark

Abstract

The Internet of Things (IoT) is becoming increasingly prevalent in today's society. Innovations in storage and processing methodologies enable the processing of large amounts of data in a scalable manner, and generation of insights in near real-time. Data from IoT are typically time-series data but they may also have a strong spatial correlation. In addition, many time-series data are deployed in industries that still place the data in inappropriate relational databases. Many open-source time-series databases exist today with inspiring features in terms of storage, analytic representation, and visualization. Finding an efficient method to migrate data into a time-series database is the first objective of the thesis. In recent decades, machine learning has become one of the backbones of data innovation. With the constantly expanding amounts of information available, there is good reason to expect that smart data analysis will become more pervasive as an essential element for innovative progress. Methods for modeling time-series data in machine learning and migrating time-series data from a database to a big data machine learning framework, such as Apache Spark, is explored in this thesis.