Computer tests as an instrument for effectiveness investigation of modeling algorithms. Yefimenko, S.M., Stepashko, V.S. IWIM, Prague, 2007.
Article (in pdf)
Abstract. Technology of testing algorithms effectiveness for structural models identification with the use of statistical computer tests is developed. The recurrent bordering algorithm is investigated. Recurrent modifications of well known Gauss and Gram-Schmidt methods are developed. The comparative testing of recurrent methods for solving of linear equations system in the parameters estimation problem as well as structural identification criteria of model construction is carried out. Effectiveness of parallel computation are explored with the purpose of extension of modeling possibilities from statistical data. Real world problems are solved for modeling of the ferromolybdenum market price and upper sedimentary layer density of the Caspian Sea bottom.
Keywords. Computer tests, modeling from data, recurrent parameter estimation, bordering method
References.
- Seber, G.A.F. Linear Regression Analysis. John Wiley and Sons, New York – London – Sydney – Toronto, 1977
- Stepashko, V.S. Optimization and Generalization of Model Sorting-out Schemes in Algorithms for the G.M.D.H., Soviet Automatic Control, 12, 4, (1979), Pp. 28-33
- Stepashko V. S., Yefimenko S. M. On the Effectiveness of Recurrent Methods of Parameter Estimation in Macromodeling Problems//Proceedings of V of International Workshop "Computational Problems of Electrical Engineering", Jazleevets, Ukraine, August 26-29, 2003, pp.106-107.
- V. S. Stepashko, and S. N. Efimenko. Sequential Estimation of the Parameters of Regression Model //Cybernetics and Systems Analysis, Springer New York, July, 2005, Vol. 41, Num. 4, pp.631-634.
- T. I. Nizamov, S. R. Ibrahimova, R. K. Quluzade, A. I. Isayev, V. S. Stepashko, S. N. Yefimenko. Hydroacoustic monitoring of water environment //Proceedings of Third International Conference on Technical and Physical Problems in Power Engineering, Ankara, Turkey, May 29-31, 2006, pp.1108-1110.
