Evaluation of GMDH artificial neural network model for predicting the spatial distribution of the family Laelapidae (Acari, Mesostigmata) in Shahrood region, Semnan province

This study aimed to predict the population of Laelapid mites in Shahrood region using an artificial neural network. The data of this family were obtained in the year 2015. In this model, the variables sampling date, longitude and latitude as the input variables, and the population of Laelapid mites were used as the output variable. The network type used was GMDH neural network that was optimized by genetic algorithms. To evaluate the ability of GMDH neural networks to predict the distribution, statistical comparison parameters such as mean, variance, statistical distribution, and coefficient determination of linear regression between predicted values and actual values were used. Results showed that in training and test phases of GMDH neural network, there was no significant effect between variance, mean, and statistical distribution of actual values and predicted values. Our map showed the patchy distribution of these predatory mites. Maps obtained from artificial neural networks help program planners to use the pest control programs, particularly if maps coordinate with geographical conformity of each location. Therefore, control was focused on areas with decreased densities of these predatory mites.