Data-driven materials science: application of ML for predicting band gap

The conventional technique of materials design and discovery relies on experimentation and complex computational methods. The sheer number of materials present makes testing each one for specific properties and application makes the entire process time and resource intensive. This has resulted into Machine Learning (ML) to gain popularity in the domain of materials science to revolutionise and accelerate the entire process. Data-driven approaches such as these can be applied to narrow down the number of materials on which we apply the tradition methods, thus cutting down both the cost and time for development of new materials. This paper demonstrates how data-driven approach can be incorporated in materials science taking the use of band gap prediction of different materials using different machine ML algorithms. Using Random Forest Regressor, the highest R squared (r2) score of 0.685 and least root mean squared error (RMSE) of 0.87 eV were achieved.