evaluation regression techniques in prediction of tractor repair and maintenance costs

the management of machine replacement (specifically tractors) is one of the most critical factors for performing field operations on time. therefore, tractor repair and maintenance costs must be accurately predicted. this research was carried out to evaluate the use of the regression technique in predicting tractor repair and maintenance costs. the study was conducted using empirical data on 60 two-wheel drive tractors from astan ghods-e razavi agro-industry. regression analyses were carried out. the response variable was repair and maintenance cost value and the independent variable was cumulative hours of use. different second and third-order polynomial models and exponential and power models of age were examined. the results showed that the cubic regression model was the best model for predicting repair, oil and fuel costs based on the coefficient of determination and significance of regression coefficients. the performance cubic regression model performed much better in the test phase for the prediction of repair costs with a mean absolute percentage error of 3.72 in comparison to oil and fuel costs with a mean absolute percentage errors of 5.21 and 9.51, respectively.