Predicting Energy Consumption Using Machine Learning

No Thumbnail Available
Chhabra, Mayank
Issue Date
MSc in Business Analytics
Dublin Business School
Electricity is the primary source of energy all around the world, and predicting the given commodity is a substantial topic. Previous research has shown low accuracy in the given topic because of the non-linear electricity consumption pattern. Our research tries to incorporate data mining methodology- KDD to predict the power consumption based on various features. The study has used various feature Engineering techniques like F regression to find the most critical variables and build statistical and deep learning models like linear regression k-nearest neighbour and Random forest. The study has incorporated statistical models to understand the relationship between the dependent and independent variables and justify the results of the machine learning models, building a comparative study. These models are evaluated on different metrics like RMSE, MAE and MAPE, and linear regression has achieved the Minimum error and can be used for the power consumption.