6. Conclusion
In this study, saturate pressure formulated to compositional
data using an epsilon SVR model. The kernel function was RBF and
it used various optimization techniques, including GA, PSO, ICA, CS
and BA. The results of constructed model were compared with
previous developed models and following conclusions are
obtained:
1. Among different algorithm which employed for optimizing SVR,
BA has better performance.
2. Comparison between current developed models and previous
developed models reveal the superiority of optimized SVR for
estimation of saturation pressure.