Tülay Korkusuz Polat is with the Department of Industrial Engineering, Engineering Faculty, Sakarya University, Esentepe Campus, 54187, Sakarya, TURKEY (korkusuz@sakarya.edu.tr ) Seher Arslankaya is with the Department of Industrial Engineering, Engineering Faculty, Sakarya University, Esentepe Campus, 54187, Sakarya, TURKEY (corresponding author to provide phone: +90 264 295 5686 aseher@sakarya.edu.tr ) The Cost Forecasting Application in an Enterprise with Artificial Neural Networks Tülay KORKUSUZ POLAT, Seher ARSLANKAYA Abstract - Nowadays, the one of sections which is studied about is Artificial Neural Network (ANN) Models. ANN researches are related to most field like optimization, control, image processing, meaning and separating language, natural language and forecasting. The inspiration of the ANNs is the power, elasticity and sensitivity of the biological brain. ANN is the mathematical model of the nerve cells, synapse and dendrites which are the main biological components of the brain. ANN is formed from simple mathematical elements. There are two kinds of learning processes in ANN; supervised and unsupervised. In the supervised learning process, the output set necessary for each input set, and both of them form the learning set. Usually, learning is used to realize by introduced to these pairs (input/output sets) to ANN. In the learning process, firstly, the input sets are given to ANN, and the output of them are computed. Afterwards, ANN changes the weights, until the desired convergence criteria level between the computed outputs and the real outputs is proved. As a result, ANN is trained and the weights at the most suitable values. In this study, the existing cost information of the factory was provided as an input for the artificial neural network, and the network was asked to yield the amount of production as an output. The study deals with an implementation of artificial neural network to determine the amount of production using the cost information obtained from the firm. Index Terms - Artificial Neural Network, Back Propagation, Cost Forecasting I. INTRODUCTION When it is intended to convert the relationships even between the simplest events of everyday life to mathematical expressions, the functions we encounter are mostly nonlinear, that is to say, exponential expressions. When we consider the same thing in terms of the examination of a process, relationship between maintenance and production, rotating, obtaining the optimum result with what is available, estimation of cost, budget planning and etc. at an enterprise, it mostly seems impossible to solve the activities at enterprises by converting them to mathematical expressions. Many applications that are based on artificial intelligence have been developed in order to solve similar relationships today, and some approaches have been developed by being inspired by the events in nature, in order to predict the consequences of relationships which cannot be converted to mathematical expressions and which cannot be solved with known algorithms (exponential expressions). The primary approaches used frequently by enterprises particularly these days include artificial neural networks, genetic algorithms, expert systems, fuzzy logic and agent, and in most of them, a heuristic approach is used to solve the problem. The approach of artificial neural networks, which arose from these necessities, has begun to be efficiently applied at enterprises, and it is used in applications with respect to issues such as optimization, tabulation, estimation of demands and estimation of cost. II. ARTIFICIAL NEURAL NETWORKS Artificial intelligence (AI) is a broad field, and means different things to different people. It is concerned with getting computers to the tasks that require human intelligence. However, having said that, there are many tasks which we might reasonably think require intelligence – such as complex arithmetic – which computer can do very easily. Conversely, there are many tasks that people do without even thinking – such as recognizing a face – which are extremely difficult to automated. AI is concerned with these difficult tasks, which require complex and sophisticated reasoning processes and knowledge (Cawsey, 1998). AI may be defined as the branch of computer science that is concerned with the automation of intelligent behavior. However, this definition suffers from the fact that intelligence itself is not very well defined or understood (Luger, 2002). AI is an area of computer science concerned designing intelligent computer systems. Artificial intelligence, like most engineering disciplines, must justify itself to the world of commerce by providing solutions to practical problems. The human brain no doubt is a highly complex structure viewed as a massive, highly interconnected network of simple processing elements called neurons. However, the behavior of a neuron can be captured are a simple model as shown in figure 1. Every component of the model bears a direct analogy to the actual constituents of a biological neuron and hence is termed as artificial neuron. It is this model which forms the basis of artificial neural networks (Rajasekaran and Vijayalakshmi Pai, 2005). Proceedings of the World Congress on Engineering 2010 Vol III WCE 2010, June 30 - July 2, 2010, London, U.K. ISBN: 978-988-18210-8-9 ISSN: 2078-0958 (Print); ISSN: 2078-0966 (Online) WCE 2010