Estimation of electric consumption of the firms operating in different sectors with machine learning algorithms

Firms or companies need to be able to draw a sustainable and effective roadmap and to estimate the electricity consumption amounts at the best level. The more successful the electricity consumption estimation is, the lower the cost of the companies. That's why it's so important in today's world. Accurate consumption estimating real-time imbalances will provide a competitive advantage with other retail sales companies. In this study, Machine Learning (ML) methods are proposed to make short-term consumption estimation and to test the accuracy of the methods, hourly consumption estimations of firms (companies) from various sectors in Turkey are made and the results are evaluated. Even if different methods perform well at different times, they cannot consistently perform the same. In this thesis study, we evaluate the results of mean absolute error (as performance measurement indicator) using Extreme Gradient Boosting Regression (XGBR) and Gradient Boosting Regression (GBR) methods and interpreted the results. The programming codes are written in Python programming language. It is observed that it is a suitable to use ML methods for the estimation of electricity consumption for future periods and similar studies.