Smart Temperature Management in Buildings using Predictive Analysis by Machine Learning Algorithms

In the contemporary world, energy management stands as a critical environmental imperative, necessitating a balanced approach that addresses both ecological and economic considerations. In the scope of this research paper, we employ machine learning techniques to tackle a specific facet of this challenge: temperature and energy management within buildings. Our primary objective is to predict the heating and cooling loads of buildings using different regression models, namely linear regression, decision tree, and random forest. We examine the performance of our models with open-source data from Kaggle-based performance metrics such as Mean Absolute Error, Mean Squared Error, and R-squared. Our findings demonstrate that all three machine learning algorithms perform effectively in predicting Heating and Cooling Loads. Notably, the decision tree regressor and random forest regressor surpass the linear regression model, exhibiting superior accuracy, as indicated by their lower Mean Absolute Error and Mean Squared Error, along with higher R-squared values. Consequently, the decision tree-based algorithm emerges as the most suitable choice for this particular dataset.