In the present work, the quantitative structure–retention relationship (QSRR) was used to predict the gas chromatographic retention factors of some organic nucleuphile on chemically modified stationary phase by complexes of Cu (II) with amino groups. The gravitation index, relative negative charge surface area, C component of moment of inertia and weighted negative charged partial surface area are selected as the most relevant descriptors from the pool of descriptors. These descriptors were used for developing multiple linear regression (MLR) and artificial neural network (ANN) models as linear and nonlinear feature mapping techniques. The root mean square errors (RMES) in calculation of retention factors for training, internal and external test set are 0.242, 0.295, and 0.240, respectively for MLR model, and for ANN model the RMSE for training, internal and external test set are; 0.084, 0.108, and 0.176. The ANN and MLR model were further examined by cross validation test, which obtained statistics of Q2 = 0.82 and SPRESS = 0.22 for MLR model and Q2 = 0.97, SPRESS = 0.07 for ANN model. Comparison between these results and other statistics of ANN and MLR models revealed the superiority of ANN over MLR model
