Data Analytics of Energy Data

Abstract: The thesis concerns mainly the construction of a pipeline that enables the analyzing,visualizing, and forecasting of time-series data in an intuitive and streamlined process.The main data set consists of four measurements: purchased heating, purchasedelectricity, and purchased cold water from select buildings at the Lule ̊a Universityof Technology campus as well as the outside temperature of the campus. This isused to establish a proof of concept, demonstrating the validity of this pipeline andits subsystems. Using this pipeline you are able to upload data, visualize data andanalyze data online and also create future forecasts of these measurements which arealso displayed online.As the global demand for energy efficiency increases, tools, like this one, is moreimportant than ever in order to give the decision-makers more insight. In the caseof the campus buildings, you might be able to more easily identify anomalous valueswhich point to some oversight that can then be amended. For instance, if two identicalbuildings exist and one of them consumes 50% more heating, you can conclude thata problem exists and now you know where the problem lies so you can amend theissue.Forecasting future consumption is also helpful since it would allow you to reducethe purchasing of fossil fuels, such as gas, which is the case at La Trobe Universityin Melbourne. Using forecasting they can better predict how much gas they need topurchase and when the peak consumption hours are so that they can adjust their solarproduction accordingly. Thus, forecasting future consumption can further reduce theglobal need and impact of fossil fuels.To conclude, this pipeline can be used as a tool to reduce the environmental impactof the Lule ̊a University of Technology campus buildings. The pipeline can then beapplied to other areas to help them solve their problems. Some of the findings ofthis thesis include comparisons of common forecasting algorithms and the benefits ofusing weekday/weekend models.In the future, this might also inspire others to make similar projects, just like LaTrobe University inspired us at the Lule ̊a University of Technology.