Classification of crystallographic materials through machine learning

The prediction of new crystallographic materials through machine learning has allowed the save of resources in the synthesis of new compounds. The search of this material with optical properties allows to reach applications in areas such as the medicine, engineering, informatics, materials, etc. Two features from the crystallographic planes were used to predict materials, Imax and 2-theta angle. The plasmons are produced by the metals and are detected when a beam of light ultraviolet influence in the surface and through the response the plasmon is described in the features absorbance and wavelength. The method to predict is the nearest neighbor rule 1-NN (1-Nearest Neighbor) that use the Euclidean distance, the algorithm can predict several neighbors, but the best choice will be the compound that presented the plasmon with more similarity. The results show that 84% of precision is achieved for predicting metal oxides with similar optical properties by machine learning method.