In this research work, the combination of the data-driven approaches based on the machine learning method and the group interactions modeling is used to predict the thermomechanical and mechanical properties of the vast number of polymers. The macroscopic properties of polymers were predicted only by using the information of their structure. For this purpose, the Random Forest (RF) algorithm as a machine learning method was used. Glass transition temperature, heat capacity, Debye temperature, thermal expansion coefficient, and bulk modulus of the polymers are the calculated properties by the mentioned methods. The performance of the models created with the machine learning method was reliable, and the models can predict with high accuracy. A good agreement between the calculated and measured heat capacity by the Modulated differential scanning calorimetry (MDSC) of the polymers was observed.