where yj is the jth response to be predicted using the (predictor)
variables, zj1 to zjr given as input. r is the number of predictor variables
and b the regression coefficients. ej is the jth residual or error
between the predicted response and the observation. The observations
are the optimised thermal efficiencies.
The least squares principle was used to determine the regression
coefficients. The method determines the coefficients that produce the minimum sum of squared residual values, i.e. the best fitted
regression line. Non-linear models were also investigated, but the
linear model showed to provide the best fit with the observed data
in all four cases.