Prediction of emission energy of Cr3+ in oxides based on first-principles calculations and machine learning