and total tested weight were very low. It could be attributed
to be small ratio of differences between the replications
(different tractors) and large quantity of these
variables which lead to be considered as the less effective
parameter. Applying these two variables caused to increase
of 1.4% and 2.93% in coefficient of determination
and accuracy of estimation in ANN model compared to
regression model. Further it can be found that the variables
of drawbar power and PTO power have the most effect on
estimating tractor fuel consumption.