Prediction of John Deere tractor repair and maintenance costs by using two different structures of MLP artificial neural network

Replacement of tractor is one of the important decisions that must be made with respect to farm machinery manager. Accurate forecasting of repair and maintenance cost is one of the most critical factors for making this decision. The purpose of this research was to evaluate the capability of two structures of MLP neural network in predicting repair and maintenance costs. First, networks were designed in order to predict the components of repair and maintenance costs individually, and then a single network was designed for simultaneous prediction of all components costs. The study was conducted using actual data on 28 John Deer tractors from Astan Ghodse Razavi agro-industry. According to the obtained results, the two types of neural networks are accurately able to predict the repair and maintenance costs. Also, the prediction of repair and maintenance cost components of tractor with a single network produces a better result than using separate networks for prediction of each cost component. Therefore, neural network can be improved the economic decision making capabilities of machinery managers.