Comparative study of support vector regression model under different kernel functions
- 1Nigerian Institute of Medical Research, Lagos, Nigeria
Res. J. Mathematical & Statistical Sci., Volume 8, Issue (1), Pages 1-5, January,12 (2020)
In spite of the fast growth in computer experiments, some computer models remain complex and often too time-consuming to be applied directly to mimic the experimentation conducted in the laboratory and predict the response of computer models using entirely a new set of data. To circumvent these problems and simplify the burden in using a simulator, a metamodel named as a support vector regression (SVR) model was used in this study as an emulator of a simple pendulum computer (SPC) model. The efficiencies of the SVR model under Gaussian Radial Basis (GRB) and B-Splines (BS) Kernel functions were examined. A SPC experiment developed using Orthogonal Array (OA)-based Latin Hypercube Design (LHD) was adopted to demonstrate the goodness of the metamodel. Further performance of SVR model at predicting the stoppage time of a pendulum bob was checked using the two kernel functions. Comparisons were also made on the performances of SVR model using R-square, relative average absolute error (RAAE) and relative maximum absolute error (RMAE), respectively. Estimated values for RAAE (0.0458) and RMAE (0.0500) provided by B-Splines functions are relatively smaller than the estimated values for RAAE (0.0604) and RMAE (0.2686) provided by the GRB function. Further results also showed that the SVR model with GRB kernel function has the estimated R2(0.9936) that is relatively smaller than the one provided by the B-Splines function (0.9977). This study concluded that the SVR model with B-Splines kernel function outperforms the one that uses Gaussian Radial Basis function for modelling and predicting the stoppage time of a pendulum bob at untried inputs. The model developments and analyses were performed using 184.108.40.2061360 (R2016a) MATLAB software.
- Osuolale K.A., Yahya W.B. and Adeleke B.L. (2017)., An enhanced algorithm for constructing optimal space-filling designs using hadamard matrices of orders 4λ and 8λ., Annals, Computer Science Series, 15(2), 9-17.
- Osuolale K.A. and Yahya W.B. (2017)., developing a javelin throw as a computer experiment using orthogonal array Latin hypercube design., Proceedings of the 1st International Conference of the Nigeria Statistical Society, Nigeria.
- Osuolale K.A., Yahya W.B. and Adeleke B.L. (2014)., Performing a simple pendulum experiment as a demonstrative example for computer experiments., Annals. Computer Science Series, 12, 33-38.
- Strogatz S. (2003)., The real scientific hero of 1953., New York Times. March 4, Editorial/Op-Ed.
- Osuolale K.A., Yahya W.B. and Adeleke B.L. (2015)., Construction of orthogonal array-based Latin hypercube designs for deterministic computer experiments., Annals. Computer Science Series, 13(1), 24-29.
- Gunn S.R. (1998)., Support vector machines for classification and regression., Technical Report, University of Southampton.
- Osuolale K.A., Yahya W.B. and Adeleke B.L. (2014)., Construction of space-filling designs for three input variables computer experiments., WASET, International Journal of Computer, Control, Quantum and Information Engineering, 8(9), 1624-1628.
- Osuolale K.A., Yahya W.B. and Adeleke B.L. (2015)., Modelling and analysis of simple pendulum computer experiment., Proceedings of the International Conference on Computer Science Research and Innovations (CoSRI), University of Ibadan.
- Osuolale K.A.(2018)., Modelling and analysis of simple pendulum computer experiment using a support vector regression model with B-Splines Kernel Function., Proceedings of the International Conference on Applied Physics and Mathematics, Dubai.
- Osuolale K.A. (2019)., Modelling and analysis of simple pendulum computer experiment using a support vector regression model., Thailand Statistician (in press).
- MATLAB Basics (2003)., Matlab/Simulink tutorial (2nd Ed.)., School of Electrical, Electronic and Computer Engineering, University of Newcastle Upon Tyne.
- Smola A.J. and Schölkopf B. (2004)., A tutorial on support vector regression., Statistics and computing, 14(3), 199-222.
- Sahama T. (2003)., Some practical issues in the design and analysis of computer experiments (Ph.D. Thesis)., Victoria University of Technology.