“…Since R value does not change by equal shifting of the output values predicted by a model, it is acknowledged not to be a good indicator on its own for evaluating the accuracy of a model on its own. On the other hand, besides assuming the impact of various data divisions for the learning and validation data, the performance index ( ρ ) simultaneously takes into account the changes of RRMSE and R. Lower RRMSE and higher R values yield in lower OBJ indicating a more accurate model (Gandomi, Roke 2015). The values obtained for R, RRMSE, and ρ are respectively, 0.8467, 0.4065, and 0.220.…”