Forecasting Cyberloafing Behavior among Youth and Sport Office Employees in Yazd Province using Artificial Neural Network Technique

The aim of this research was Investigating and Prioritizing Factors Affecting Cyberloafing Behavior among Youth and Sport Office Employees in Yazd Province using Artificial Neural Network Technique. This study was an applied research in terms of purpose and a descriptive- survey research in terms of method. The statistical population was the employees of youth and sport office in Yazd province. The statistical sample was selected using simple random sampling method. For gathering data about dependent variable, 22-item cyberloafing behavior scale by Stoddart (2016) was used. Reliability of scale was confirmed by Cronbach’s Alfa and validity of the scale was confirmed by content analysis. Data were analyzed using SPSS software version 24. For analysis of data including 23 independent variables and one dependent variable, two types of neural network including MLP and RBF were designed and implemented. Correct percent of cyberloafing prediction in the training, testing and validation data for the MLP neural network was 87.8, 75.0 and 72.7, respectively, and 83.6, 81.8 and 91.7 for the RBF neural network, respectively. The area under the rock for MLP and RBF networks was 0.611 and 0.677 respectively. Comparison of two MLP and RBF neural networks based on rock curve and prediction Correct percent showed that RBF neural network is more effective in forecasting cyberloafing, and the variables Proximity of supervisor, teaching master students, age and locus of control had the greatest impact on cyberloafing.