By Costa, C.; Santos, M.Y.
Lecture Notes in Engineering and Computer Science
Nowadays, cities consume more energy to fuel their day-to-day activities. With the rise of electrical devices we face more challenges associated with energy control and distribution. Apart from this, we also spend a lot of energy trying to either heating or cooling our homes. This paper illustrates an architecture to extract, load, transform, mine and forecast Big Data. This technological architecture makes use of a dataset containing electricity and gas consumption of homes distributed within multiple USA cities and states. The main purpose of our work consists in delivering to citizens a new form of self-monitoring their electricity and gas consumption, by comparing them to other homes within their cluster or state and by forecasting future energy consumptions. Moreover, the architecture also delivers to energy providers and cities a smarter overview of the energy landscape. This work uses simulated data from United States of America along with Hadoop, WEKA and Tableau to store and process Big Data, to produce clusters and time series forecasts, and to visualize information, respectively. The results reveal that, using this architecture, it is possible to produce accurate clusters of homes based on their energy consumption and it is also possible to forecast future electricity consumptions with a small margin of error.