Accurate and fast design cooling load calculation for air-conditioning systems is essential for optimal design and energy-efficient operation. A data-driven model based on explainable feature selection (EFS) and multi-task learning (MTL) was proposed for the calculation of design cooling loads under different non-guarantee rates. However, the superiority of the EFS-MTL model to other machine learning models was not discussed. Therefore, in this research, a comparative study was conducted to illustrate that the EFS-MTL has better performance comparing to other machine learning model. The results demonstrates that the EFS-MTL model shows better MAE and MAPE to those of other machine learning models.