Abstract
The statistical properties of cross-validation estimation as a criterion for choosing a decision model (a method for generating the decision function) are studied in the paper. For the variance analysis problem it is proved that the cross-validation criterion is equivalent to Fisher’s criterion for testing a homogeneity hypothesis under a certain significance level. It is revealed that the cross-validation criterion used for choosing the decision function among a certain one-parameter class and optimal decision function generation in the framework of a Bayesian model with normal parameter distribution are the same.
Similar content being viewed by others
References
V. B. Berikov and G. S. Lbov, “Choice of optimal complexity for logical decision functions in pattern recognition problems”, Dokl. Math. 76, 969–971 (2007).
V. B. Berikov and I. A. Pestunov, “Creating a cluster ensemble for hyperspectral images segmentation”, Vychisl. Tekhnol. (Comput. Technologies) 21 (1), 15–24 (2016) [in Russian].
I. E. Genrikhov, E. V. Djukova, and V. I. Zhuravlyov, “About full regression decision trees”, Mashinnoe Obuchen. Anal. Dannykh (Mach. Learn. Data Anal.) 2 (1), 116–126 (2016) [in Russian]. doi: https://doi.org/10.21469/22233792.2.1.09
N. G. Zagoruiko and G. S. Lbov, “The choice problem in data analysis and control theory”, Sib. Zh. Ind. Mat. 3 (1), 101–109 (2000) [in Russian].
V. M. Nedel’ko, “Some aspects of estimating a quality of decision functions construction methods”, Vest. Tomsk. Gos. Univ., Ser. Upr., Vych. Tekh. Inform. No. 3 (24), 123–132 (2013) [in Russian].
K. V. Vorontsov, “Combinatorial bounds for learning performance”, Dokl. Math. 69, 145–148 (2004).
P. A. Turkov, O. V. Krasotkina, and V. V. Mottl, “Feature selection in the classification problem under concept drift”, Izv. Tul. Gos. Univ., Ser. Estestv. Nauki, No. 4, 67–78 (2015) [in Russian].
A. P. Kovalevskii and E. V. Shatalin, “The choice of a regression model of the body weight on the height via an empirical bridge”, Vest. Tomsk. Gos. Univ., Ser. Mat. Mekh. No. 5 (37), 35–47 (2015) [in Russian].
M. Yu. Khachai and M. I. Poberii, “Scheme of boosting in the problems of combinatorial optimization induced by the collective training algorithms”, Autom. Remote Control 75, 81–93 (2014).
Yu. Yu. Linke, “Asymptotic properties of one-step weighted M-estimators with application to regression problems”, Teor. Veroyatn. Primen. (Theory Probab. Its Appl.) 62 (3), 468–498 (2017) [in Russian].
V. M. Nedel’ko, “Estimation of feature importance for quantile regression”, Mashinnoe Obuchen. Anal. Dannykh (Mach. Learn. Data Anal.) 3 (2), 151–159 (2017) [in Russian]. doi: 10.21469/22233792.3.2.05
V. M. Nedel’ko, “Regression models in a classification problem”, Sib. Zh. Ind. Mat. 17 (1), 86–98 (2014) [in Russian].
Vl. D. Mazurov and M. Yu. Khachai, “Boosting and the polynomial approximability of the problem on a minimum affine separating committee,” Trudy Inst. Mat. Mekh. UrO RAN 19 (2), 231–236 (2013) [in Russian].
N. Spirin and K. Vorontsov, “Learning to rank with nonlinear monotonic ensemble,” in Multiple Classifier Systems, Proc. 10th International Workshop, MCS 2011, Naples, Italy, June 15–17, 2011, Ed. by C. Sansone, J. Kittler, and F. Roli, Lecture Notes on Computer Science (Springer, Berlin, 2011), Vol. 6713, pp. 16–25.
Author information
Authors and Affiliations
Corresponding author
Additional information
Viktor Mikhailovich Nedel’ko, born in 1971, graduated from Novosibirsk State University in 1993. Candidate of mathematical and physical sciences, associate professor, Sobolev Institute of Mathematics of Siberian Branch of the Russian Academy of Sciences. Fields of interests: mathematical methods of data mining (machine learning), mathematical statistics, artificial intelligence. Publications: more than 60 papers.
Rights and permissions
About this article
Cite this article
Nedel’ko, V.M. Statistical Fitting Criterion on the Basis of Cross-Validation Estimation. Pattern Recognit. Image Anal. 28, 510–515 (2018). https://doi.org/10.1134/S1054661818030148
Received:
Published:
Issue Date:
DOI: https://doi.org/10.1134/S1054661818030148