Predicting thermal conductivity of nanomaterials by correlation weighting technological attributes codes

A number of characteristics that include atom compositions, conditions of synthesis and the features of nanomaterials related to their commercial manufacturing have been examined as possible descriptors of a given nanostructure. Using an optimization procedure linked to the Monte Carlo method the special correlation weights have been calculated for each descriptor. A new application of the correlation weights predictive model for the thermal conductivity of nanomaterials has been developed. Statistical characteristics of the model are as follows: n = 43, r2 = 0.8687, s = 5.14 W/m/K, F = 271 (training set); n = 15, r2 = 0.8598, s = 4.91 W/m/K, F = 80 (test set).