Comparison of services quality assessment models between students from teaching and learning process quality by using artificial neural network

the aim of this study is quantity determining and evaluating of the quality position of teaching and learning process. So the artificial neural networks were used for modeling nonlinear relationships to inspect and evaluate different models of service quality evaluation. The statistical population was including 1070 peoples, which were master and PhD students of Faculty of Agriculture of Ferdowsi University (Mashhad). 280 questionnaires were collected by using Morgan table; from which 202 questionnaires were finally analyzed. In order to inspect and evaluate the quality of teaching and learning process, four models were used with artificial neural networks; including non-weighted Servprof, non-weighted Servqual, weighted Servprof and weighted Servqual. The results of the artificial neural network method showed that weighted Servqual model is more accurate to evaluate the quality of teaching and to predict satisfactory. 7-29-14-1 architecture with 29 and 14 neurons respectively in the first and second hidden layers and one neuron (weighted Servqual) in the output layer was chosen as the best model for determining the quality evaluation. This architecture has the best results for R (0.96), MAE(0/18), MSE (0/06) and MAPE (4.41%) between the actual and modeled values.