Energy Consumption Forecasting in Buildings based on heating and cooling loads using Regression models

The Energy consumption in buildings is a significant factor in both cost-effectiveness and environmental sustainability. To optimize energy use and sustain heating, cooling, and air conditioning (HVAC) in the buildings, predictive models that can effectively estimate heating and cooling loads are crucial. This paper employs a variety of regression techniques to develop prediction models for predicting Cooling Loads (CL) Heating Loads (HL) and in buildings. XG Boost, Ridge regression, Lasso regression, Decision Tree, Random Forest, and Bayesian Ridge Regression are a few ML models used. Numerous building metrics that are connected to energy are included in the collection. The models are trained, tested, and optimized using metrics like R2 score, MAE-Mean Absolute Error, MSE-Mean Squared Error, and RMSE-Root Mean Squared Error. To improve its capacity for prediction, the model's hyperparameters are also created using GridSearchCV. Analysis of the advantages and disadvantages of each model is feasible thanks to the results' systematic comparison and visualization. The results underline how crucial it is to choose models carefully and accurately to increase energy efficiency and inform building designs. This study offers useful tips on how to use ensemble techniques and machine learning techniques to evaluate power usage in the buildings.