Intelligence approach in improving business processes

Abstract

Abstract. Nowadays, business intelligence (BI) has the top-most priority for the contemporary enterprises. The aim of this paper is to emphasize the advantages of this computer based approach in improving business processes. Data analytics and modeling of sales prediction system for enterprise is realized with artificial neural networks (ANNs). For the purpose of this research are created over 100 artificial neural networks. Three types of artificial neural networks are designed and evaluated: Function fitting neural networks, Focused Time-delay neural networks and NARX (Non-linear autoregressive) neural networks. Also were performed simulations with different architectures that differ to the number of delays and the number of neurons in the hidden layer. The network prediction performance was evaluated with Average Percentage Error (APE) and Root Mean Square Error (RMSE). The obtained results show big accuracy in prediction of the product sale. The precise prediction has influence to optimization of most of the business processes such as: supply of raw materials, organization of production process, staff scheduling, plan the electricity demand, cost reduction etc.