Prediction of Electron Band Gap of $\mathrm{A}_{2}\text{XY}_{6}$ Perovskite Compounds using Machine Learning

Increasing population and industrialization haveled to an uptick in energy requirements. Many traditional energy sources are not anymore attractive due to climate change, instead, the interest has turned to power generation from renewable sources, such as wind energy, hydro-power, and solar energy. The wide availability of sunlight and simplicity in converting sunlight to electricity has led to the search for synthesized semiconductors that give high efficiency in this conversion. A family of such semiconductors attains the perovskite structure, the most established being Methyl Ammonium Lead Iodide. The shortcomings of this compound include lead poisoning, motivating the search for perovskite structures that have low electron band-gap and are stable. A family of such perovskite structures is compounds that attain an
$\mathbf{A}_{2}\mathbf{XY}_{6}$
type structure. This paper demonstrates some methods that can be used to calculate the electron band-gap of such compounds. The metrics found from Support Vector Machine Regression and Random Forest Regression are compared and analyzed to propose a scalable model for predicting electron band-gap.