Simulating and forecasting energy consumption plays an important role in setting energy policy and making decisions towards sustainable development. This study uses statistical techniques and artificial intelligence tools including neural networks, vector support (SVM - Support vector machine), classification and regression (CART), anise Linear regression (LR-linear regression), general linear regression, automatic Chi-squared interaction detection and composite model to predict energy consumption in apartment buildings.

The data set to build a model of 200 samples, was surveyed in many apartments in Ho Chi Minh City. The single model that has the best effect in the prediction process is CART, while the best synthesized model is CART and GENLIN.