Also, in Table 4 was given two performance indices of
stepwise multiple range regression models and ANN model
(prediction accuracy and coefficient of determination).
From Table 4 can be inferred that the stepwise regression
model with additional input variables approximately represented
an increase in accuracy at the step of 1 to 2 while after
step of 3 its accuracy were approximately unchanged. These
observations can be explained as follows: with increasing
the number of independent variables influenced on tractor
fuel consumption, the ability of stepwise multiple ranges
model in interpretation of the complicated relationships
among influencing variables limited and this model is not
capable to solve the multifaceted relations between given
variables while the neural network model is capable to learn
the complex relationships among variables very well and
with increasing in input variables its accuracy improved.