Abstarct: In this study, quantitative structure-property (QSPR) models were used to predict the retention index of some essential oils of the leaves of Pyronium tenchinense using different variable selection and modeling methods. Variable selection methods of Gray Wolf (GWO) and Whale (WOA) optimization algorithms were used to select the best descxriptors, and Support Vector Machine (SVM) and Artificial Neural Network (ANN) were used as nonlinear modeling methods. The results of the SW-ANN, GWO-ANN and WOA-ANN baxsed models showed that the GWO-ANN model performed better with a higher coefficient of determination and lower error. Also, only three descxriptors appeared in this model, which is an advantage for this model. The modeling results using the support vector machine method and using two variable selection methods WOA and GWO showed that the WOA-SVM model with a lower external test RMSE was significantly superior to the GWO-SVM model. This superiority indicates that the Whale optimization algorithm (WOA) was more successful in selecting important descxriptors for modeling the retention index of essential oils of Pyronium tenchinense plant. Overall, the combination of the new Wolf and Whale algorithms in combination with artificial neural networks and support vector machines provides efficient methods for predicting the retention index of these compounds.