Abstract:

The artificial nerve network (YSA), one of the artificial intelligence technologies, is widely used in many social and engineering fields such as prediction, modeling, classification and so on. The parallel structure of YSA is an important feature for real-time applications and its important advantages are the flexible and compatible nature of the model. YSAs can be easily used for any application without having to reprogram once they have been trained. YSAs use the appropriate learning method and determine the linear and non-linear relationships between output patterns. In other words, they have the ability to model systems using the correlation between entrance and output of complex and nonlinear systems. In this study, the time of conversation is expected to be done with the number of calls expected to come to the YSA method and call center data. Call centers are requests, opinions, suggestions, dissatisfaction, complaints, etc. from customers/city citizens of the institutions. It is a high level of communication that it serves. The call center managers are faced with the possibility of making predictions in early decision making. Therefore, it is important to estimate the time of calls received with the number of calls received in daily, weekly and monthly periods in the call center systems’ reports. In this study, the number of interviews and interviews from the previous months were used as training data. For predictive modeling, the advanced powered YSA model has been obtained from entrance to exit and the network model has been trained with the Levenberg-Marquardt algorithm. Linear activation functions have been used for the hidden and output layers of this three-layer YSA, which consists of the input, hidden and output layers. The number of calls that come with the advanced powered and recovered YSA model used in this study and the time of calls were predicted and obtained and the predictive performance of this YSA model was observed that this model was reliable and consistent.