The R 2 and RE for training and test selleck chemicals llc sets were (0.861, 0.748) and (14.37, 23.09),
respectively. For the constructed model, two general statistical parameters were selected to evaluate the prediction ability of the model for the log (1/EC50). The predicted selleck chemical values of log (1/EC50) are plotted against the experimental values for training and test sets in Fig. 5. Consequently, as a result, the number of components (latent variables) is less than the number of independent variables in KPLS analysis. The statistical parameters highest square correlation coefficient leave-group-out cross validation (R 2) and relative error
(RE) were obtained for proposed models. Each of the statistical parameters mentioned above was used for assessing Adriamycin in vivo the statistical significance of the QSAR model. This GA-KPLS approach currently constitutes the most accurate method for predicting the anti-HIV biological activity of the drug compounds. The KPLS model uses higher number of descriptors that allows the model to extract better structural information from descriptors to result in a lower prediction error. This suggests that GA-KPLS holds promise for applications in choosing variables for L–M ANN systems. This result indicates that the log (1/EC50) of these drugs possesses some nonlinear characteristics. Fig. 5 Plots of predicted log (1/EC50) against the experimental values by GA-KPLS model Abiraterone Results of the L–M ANN model With the aim of improving the predictive performance of nonlinear QSAR model, L–M ANN modeling was performed. The networks were generated using the 14 descriptors appearing in
the GA-KPLS models as their inputs and log (1/EC50) as their output. For ANN generation, data set was separated into three groups: calibration, prediction, and test sets. A three-layer network with a sigmoid transfer function was designed for each ANN. Before training the networks, the input and output values were normalized between −1 and 1. Then, the network was trained using the training set and the back propagation strategy for optimizing the weights and bias values. The proper number of nodes in the hidden layer was determined by training the network with different number of nodes in the hidden layer. The root-mean-square error (RMSE) value measures how good the outputs are in comparison with the target values.