Abstarct: Activity coefficient at infinite dilution (γ∞) is a key parameter which can be used for the selection of effective solvent in the separation processes. In this work, a quantitative structure-property relationship (QSPR) approach was employed to predict activity coefficients at infinite dilution for 60 organic compounds and water in the ionic liquid 1-butyl-1-methylmorpholinium tricyanomethanide ([BMMOR][TCM]) at six different temperature. A large number of descxriptors were calculated using HyperChem and Dragon software and the best calculated descxriptors was selected by stepwise regression (SR) and combined data splitting-feature selection (CDFS) strategy. A total of 12 molecular descxriptors, as the most feasible ones, were selected using SR method. Also, 7 significant molecular descxriptors and one experimental variable (temperature) were selected using CDFS method. The selected descxriptors by two methods were used as input for artificial neural network (ANN) and support vector machine (SVM). After training and optimization of the ANN and SVM parameters, the performance of each model was evaluated using the test set. The mean square error (MSE) value for the test data set from SR-ANN, CDFS-ANN, SR-SVM and CDFS-SVM models were 0.0515, 0.1322, 0.0836 and 0.0988, respectively. The obtained results showed the superiority of the SR-ANN than other methods.