Choice of mathematical model: balance between complexity and proximity to measurements

A.V. Sokolov, V.V. Voloshinov

Abstract


Mathematical model selection method on the basis of a balance between the complexity and the experimental data fitting accuracy is proposed. The method is based on: 1) a set of (parametric) families of models suitable for satisfactory reproduction of measurements; 2) model complexity notion formalization (for the selected family of models); 3) a cross-validation procedure for estimating the error of data modeling; 4) search for the optimal trade-off between the complexity of the model and proximity to measurements based on minimizing the cross-validation error of data modeling. The procedure is explained by the demonstrative case study. General mathematical statement of model evaluation problem is presented. The issues of software implementation in a distributed computing environment are discussed.

Full Text:

PDF (Russian)

References


A. Makroglou, J. Li, Y. Kuang. “Mathematical models and software tools for the glucose-insulin regulatory system and diabetes: an overview”, in Proceedings of the 2005 IMACS, pp. 561 – 565

Gomenjuk S.M., Emel'janov A.O., professor, d.f.-m.n. Karpenko A.P., Chernecov S.A. “Metody prognozirovanija optimal'nyh doz insulina dlja bol'nyh saharnym diabetom I tipa. Obzor”, Jelektronnoe izdanie “NAUKA i OBRAZOVANIE”. Izdatel' FGBOU VPO "MGTU im. N.Je. Baumana". Jel # FS 77 - 48211. ISSN 1994-0408

Tihonov A.N. “O matematicheskih metodah avtomatizacii obrabotki nabljudenij”, v sb. Problemy vychislitel'noj matematiki. M.: Izd-vo MGU, 1980. str 3-17.

Rozhenko, A.I. Teorija i algoritmy variacionnoj splajn-approksimacii. Novosibirsk: Izd. IVMiMG SO RAN (2005)

Hardle V. Prikladnaja neparametricheskaja regressija. M.: Mir, 1993. 349 s.

Kuhn, Max, and Kjell Johnson. “Applied predictive modeling”. Vol. 26. New York: Springer, 2013. http://appliedpredictivemodeling.com/, doi 10.1007/978-1-4614-6849-3

Hastie T., Tibshirani R., and Friedman J. Unsupervised learning. In: The elements of statistical learning. New-York: Springer, 2009. P. 485-585. DOI:10.1007/978-0-387-84858-7

Sukhoroslov, O., Volkov, S., Afanasiev, A. “A Web-Based Platform for Publication and Distributed Execution of Computing Applications”, in Parallel and Distributed Computing, 14th International Symposium on IEEE, pp. 175-184, (2015)

Samarskij A.A., Mihajlov A.P. Matematicheskoe modelirovanie. Idei. Metody. Primery. - 2-e izd., ispr. - M.: Fizmatlit, 2001. - 320 s. - ISBN 5=9221-0120-H.

Forrester, A., Sobester, A., Keane, A.: Engineering Design via Surrogate Modeling. A Practical Guide. Wiley, New York (2008), DOI 10.1002/9780470770801

Beljaev M.G., Ljubin A.D. "Osobennosti optimizacionnoj zadachi, voznikajushhej pri postroenii approksimacii mnogomernoj zavisimosti" v Tr. konf.“Informacionnye tehnologii i sistemy”(ITiS’11), 2011, s. 415--422, http://itas2011.iitp.ru/pdf/1569478557.pdf

Minu M. Matematicheskoe programmirovanie. Teorija i algoritmy. M.: Nauka (1990)

Pshenichnyj, B.N. Metod linearizacii. M.: Nauka (1983)

Hart, W.E., Laird, C., Watson, J.-P., Woodruff D.L. Pyomo–optimization modeling in python, vol. 67: Springer, 238 p., (2012)

Gay, David M. "Writing. nl files", in Optimization and Uncertainty Estimation (2005).

Smirnov, S., Voloshinov, V., Sukhosroslov, O. “Distributed Optimization on the Base of AMPL Modeling Language and Everest Platform”, in Procedia Computer Science, vol. 101, pp. 313-322 (2016)

Afanasiev A.P., Sokolov A.V., Voloshinov V.V. “Inverse Problem in the Modeling on the Basis of Regularization and Distributed Computing in the Everest Environment” in Data Analytics and Management in Data Intensive Domains: Collection of Scientific Papers of the XIX International Conference DAMDID / RCDL’2017 (October 10–13, 2017, Moscow, Russia). Eds. L. A. Kalinichenko, etc. — Moscow: FRC CSC RAS, s. 132-140, (2017) http://damdid2017.frccsc.ru/files/DAMDID_RCDL_2017_Proceedings.pdf


Refbacks

  • There are currently no refbacks.


Abava   MSU conference 2018

ISSN: 2307-8162