Newspaper Vendor Sales Prediction Using Artificial Neural Networks

Every newspaper publisher faced with the problem of determining the number of copies of newspaper and distributing them to the retail traders. Two aspects need to be balanced out in order to optimize the economical success which is the number of unsold copies should be minimal to reduce the cost of production, and the sell-out rate should be minimal to maximize the number of sold copies. Thus, a good sales rate prediction is necessary to optimize both antagonistic aspects. This paper utilized artificial neural network to predict newspaper sales for one vendor in the area of Sungai Petani, Malaysia. The predicted sales value can help the company to optimize their sales. The main objective is to develop a prototype that apply artificial neural network so that it can predict the future trend as well as the future daily sale. The network will consist of three layer which is input layer, one hidden layer and output layer. The input layer will have six input node where this will be the factor that will affect the output which is the number of copies that sold. The network will be trained with history data of a one year records of data. The output produced has the error value as low as 1.24% while the correlation coefficient between prediction and actual value is 0.1197.