procyberВестник кибернетикиProceedings in Cybernetics1999-7604Бюджетное учреждение высшего образования Ханты-Мансийского автономного округа – Югры «Сургутский государственный университет»10.34822/1999-7604-2021-4-78-82procyber-402Research ArticleФИЗИКО-МАТЕМАТИЧЕСКИЕ НАУКИPhysics and MathematicsУСТОЙЧИВОСТЬ АЛГОРИТМОВ ОТБОРА ПРИЗНАКОВ К ОШИБКАМ ВТОРОГО РОДАRESISTANCE OF THE ALGORITHMS FOR FEATURE SELECTION TO TYPE II ERRORSЧеремухинА. Д.CheremukhinA. D.

E-mail: ngieu.cheremuhin@yandex.ru

E-mail: ngieu.cheremuhin@yandex.ru

Нижегородский государственный инженерно-экономический университет, КнягининоРоссияNizhny Novgorod State University of Engineering and Economics, KnyagininoRussian Federation20212812202104 (44)7882Copyright © Черемухин А.Д., 20212021Черемухин А.Д.Cheremukhin A.D.Данная работа распространяется под лицензией Creative Commons Attribution 4.0.This work is licensed under a Creative Commons Attribution 4.0 License.https://www.vestcyber.ru/jour/article/view/402

Предметом исследования является эффективность работы алгоритмов отбора признаков применительно к задачам регрессии в контексте частоты выявления ими ложных статистически значимых зависимостей. Целью стало построение соответствующей методики и апробация ее на сгенерированных данных, а также проверка гипотезы о наличии частоты появления ошибок второго рода от распределения зависимой переменной. Проведено изучение 7 методов отбора признаков: Simulated Annealing, Select Difference, Hill-Climbing, Las Vegas, Sequential Forward Selection, Select Slope, Whale Optimization. В качестве зависимых переменных выбраны переменные, которые подчинялись 8 видам распределений (бета, Коши, экспоненциальное, гамма, логнормальное, нормальное, равномерное, Вейбулла). Установлено, что при строгом подходе к оценке качества моделей вероятность использования в практической деятельности ложнозначимых моделей невелика.

The subject of the study is the efficiency of feature selection algorithms in relation to regression problems in the context of frequency of detecting false statistically significant dependencies.The aim of the study is to build a suitable methodology and to test it on generated data, to test the hypothesis of frequency of occurrence of type II errors from distribution of dependent variable. In total, 7 methods of feature selection were studied in the work: Simulated Annealing, Select Difference,Hill-Climbing, Las Vegas, Sequential Forward Selection, Select Slope, Whale Optimization. Variables distributed according to 8 laws (Beta, Cauchy, exponential, Gamma, log-normal, normal, uniform, Weibull) were chosen as dependent variables. As a result of the study, it was found that the probability of using practical false-valued models is small using a rigorous approach in assessing the quality of models.

отбор признаковрегрессияложноположительный результатстатистическая значимость.feature selectionregressionfalse-positive resultstatistical significanceReferencesDjordjevic M., Salom I., Markovic S., Rodic A., Milicevic O., Djodrjevic M. Inferring the Main Drivers of SARS-Cov-2 Global Transmissibility by Feature Selection Methods // Geo-Health. 2021. Vol. 5, Is. 9. P. e2021GH000432.Djordjevic M., Salom I., Markovic S., Rodic A., Milicevic O., Djodrjevic M. Inferring the Main Drivers of SARS-Cov-2 Global Transmissibility by Feature Selection Methods // Geo-Health. 2021. Vol. 5, Is. 9. P. e2021GH000432.Kaliappan J., Srinivasan K., Qaisar S. M., Sundararajan K., Chang C.-Y., C S. Performance Evaluation of Regression Models for the Prediction of the COVID-19 Reproduction Rate // Front Public Health. 2021. Vol. 9. P. 729795.Kaliappan J., Srinivasan K., Qaisar S. M., Sundararajan K., Chang C.-Y., C S. Performance Evaluation of Regression Models for the Prediction of the COVID-19 Reproduction Rate // Front Public Health. 2021. Vol. 9. P. 729795.Conlon E. M., Liu X. S., Lieb J. D., Liu J. S. Integrating Regulatory Motif Discovery and Genome-Wide Expression Analysis // Proc Natl Acad Sci USA. 2003. Vol. 100, № 6. Р. 3339‒3344.Conlon E. M., Liu X. S., Lieb J. D., Liu J. S. Integrating Regulatory Motif Discovery and Genome-Wide Expression Analysis // Proc Natl Acad Sci USA. 2003. Vol. 100, № 6. Р. 3339‒3344.Zhong W., Zeng P., Ma P., Liu J. S., Zhu U. RSIR: Regularized Sliced Inverse Regression for Motif Discovery // Bioinformatics. 2005. Vol. 21, No. 22. Р. 4169‒4175.Zhong W., Zeng P., Ma P., Liu J. S., Zhu U. RSIR: Regularized Sliced Inverse Regression for Motif Discovery // Bioinformatics. 2005. Vol. 21, No. 22. Р. 4169‒4175.Shastry K. A., Sanjay H. A. A Modified Genetic Algorithm and Weighted Principal Component Analysis Based Feature Selection and Extraction Strategy in Agriculture // Knowledge-Based Systems. 2021. Vol. 232. P. 107460.Shastry K. A., Sanjay H. A. A Modified Genetic Algorithm and Weighted Principal Component Analysis Based Feature Selection and Extraction Strategy in Agriculture // Knowledge-Based Systems. 2021. Vol. 232. P. 107460.Jiménez F., García J. M., Sciavicco G., Pechuán L. M. Multi-Objective Evolutionary Feature Selection for Online Sales Forecasting // Neurocomputing. 2017. Vol. 234. P. 75‒92. URL:Jiménez F., García J. M., Sciavicco G., Pechuán L. M. Multi-Objective Evolutionary Feature Selection for Online Sales Forecasting // Neurocomputing. 2017. Vol. 234. P. 75‒92. URL:http://dx.doi.org/10.1016/j.neucom.2016.12.045 (дата обращения: 02.10.2021).http://dx.doi.org/10.1016/j.neucom.2016.12.045 (дата обращения: 02.10.2021).Blesser B. A., Kuklinski T. T., Shillman R. J. Empirical Tests for Feature Selection Based on a Psychological Theory of Character Recognition // Pattern Recognition. 1976. Vol. 8, Is. 2. P. 77–85.Blesser B. A., Kuklinski T. T., Shillman R. J. Empirical Tests for Feature Selection Based on a Psychological Theory of Character Recognition // Pattern Recognition. 1976. Vol. 8, Is. 2. P. 77–85.Tang J., Liu. H. Feature Selection for Social Media Data // ACM Trans Knowl Discov Data. 2014. Vol. 8, Is. 4. Р. 19.Tang J., Liu. H. Feature Selection for Social Media Data // ACM Trans Knowl Discov Data. 2014. Vol. 8, Is. 4. Р. 19.Gao Y., Xu A., Hu P. J.-H., Cheng T.-H. Incorporating Association Rule Networks in Feature Category-Weighted Naïve Bayes Model to Support Weaning Decision Making // Decis Support Syst. 2017. Vol. 96. Р. 27‒38.Gao Y., Xu A., Hu P. J.-H., Cheng T.-H. Incorporating Association Rule Networks in Feature Category-Weighted Naïve Bayes Model to Support Weaning Decision Making // Decis Support Syst. 2017. Vol. 96. Р. 27‒38.Yuan H., Lau R. Y. K., Xu W. The Determinants of Crowdfunding Success: A Semantic Text Analytics Approach // Decis Support Syst. 2016. Vol. 91. Р. 67‒76.Yuan H., Lau R. Y. K., Xu W. The Determinants of Crowdfunding Success: A Semantic Text Analytics Approach // Decis Support Syst. 2016. Vol. 91. Р. 67‒76.Tibshirani R. Regression Shrinkage and Selection via the Lasso // J R Statist Soc B. 1996. Vol. 58, No. 1. Р. 267‒288.Tibshirani R. Regression Shrinkage and Selection via the Lasso // J R Statist Soc B. 1996. Vol. 58, No. 1. Р. 267‒288.Fan J., Li R. New Estimation and Model Selection Procedures for Semiparametric Modeling in Longitudinal Data Analysis // J Amer Statist Assoc. Vol. 99, № 467. Р. 710‒723.Fan J., Li R. New Estimation and Model Selection Procedures for Semiparametric Modeling in Longitudinal Data Analysis // J Amer Statist Assoc. Vol. 99, № 467. Р. 710‒723.Efron B., Hastie T., Johnstone I., Tibshirani R. Least Angle Regression (with Discussion) // Ann Statist. 2004. Vol. 32, Is. 2. Р. 407‒499.Efron B., Hastie T., Johnstone I., Tibshirani R. Least Angle Regression (with Discussion) // Ann Statist. 2004. Vol. 32, Is. 2. Р. 407‒499.Khalili A. An Overview of the New Feature Selection Methods in Finite Mixture of Regression Models // JIIRS. 2011. Vol. 10, Is. 2. Р. 201‒235.Khalili A. An Overview of the New Feature Selection Methods in Finite Mixture of Regression Models // JIIRS. 2011. Vol. 10, Is. 2. Р. 201‒235.Zhang L., Mistry K., Lim C. P., Neoh S. C. Feature Selection Using Firefly Optimization for Classification and Regression Models // Decis Support Syst. 2018. Vol. 106. P. 64‒85.Zhang L., Mistry K., Lim C. P., Neoh S. C. Feature Selection Using Firefly Optimization for Classification and Regression Models // Decis Support Syst. 2018. Vol. 106. P. 64‒85.Shang R., Chang J., Jiao L., Xue Y. Unsupervised Feature Selection Based on Self-Representation Sparse Regression and Local Similarity Preserving // International Journal of MachineShang R., Chang J., Jiao L., Xue Y. Unsupervised Feature Selection Based on Self-Representation Sparse Regression and Local Similarity Preserving // International Journal of MachineLearning and Cybernetics. 2019. Vol. 10. P. 757‒770.Learning and Cybernetics. 2019. Vol. 10. P. 757‒770.Aragón-Royón F., Jiménez-Vílchez A., Arauzo-Azofra A., Benitez J. M. FSinR: An Exhaustive Package for Feature Selection // arXiv:2002.10330 [cs.LG]. 2020. URL: https://arxiv.org/ abs/2002.10330 (дата обращения: 02.10.2021).Aragón-Royón F., Jiménez-Vílchez A., Arauzo-Azofra A., Benitez J. M. FSinR: An Exhaustive Package for Feature Selection // arXiv:2002.10330 [cs.LG]. 2020. URL: https://arxiv.org/ abs/2002.10330 (дата обращения: 02.10.2021).Posario F., Thangadurai K. Simulated Annealing Algorithm for Feature Selection // International Journal of Computers & Technology. 2016. Vol. 15, № 2. Р. 6471‒6479.Posario F., Thangadurai K. Simulated Annealing Algorithm for Feature Selection // International Journal of Computers & Technology. 2016. Vol. 15, № 2. Р. 6471‒6479.Gelbart D., Morgan N., Tsymbal A. Hill-Climbing Feature Selection for Multi-Stream ASR // INTERSPEECH 2009. URL: https://www.icsi.berkeley.edu/pubs/speech/gelbart-2009.pdf (дата обращения: 02.10.2021).Gelbart D., Morgan N., Tsymbal A. Hill-Climbing Feature Selection for Multi-Stream ASR // INTERSPEECH 2009. URL: https://www.icsi.berkeley.edu/pubs/speech/gelbart-2009.pdf (дата обращения: 02.10.2021).Nandy G. An Enhanced Approach to Las Vegas Filter (LVF) Feature Selection Algorithm // 2nd National Conference on Emerging Trends and Applications in Computer Science. 2011. P. 1‒3. URL: https://ieeexplore.ieee.org/document/5751392 (дата обращения: 02.10.2021).Nandy G. An Enhanced Approach to Las Vegas Filter (LVF) Feature Selection Algorithm // 2nd National Conference on Emerging Trends and Applications in Computer Science. 2011. P. 1‒3. URL: https://ieeexplore.ieee.org/document/5751392 (дата обращения: 02.10.2021).Marcano-Cedeño A., Quintanilla J., Cortina-Januchs G., Andina D. Feature Selection Using Sequential Forward Selection and Classification Applying Artificial Metaplasticity Neural Network // 36th Annual Conference on IEEE Industrial Electronics Society. 2010. P. 2845‒2850. URL:Marcano-Cedeño A., Quintanilla J., Cortina-Januchs G., Andina D. Feature Selection Using Sequential Forward Selection and Classification Applying Artificial Metaplasticity Neural Network // 36th Annual Conference on IEEE Industrial Electronics Society. 2010. P. 2845‒2850. URL:https://ieeexplore.ieee.org/document/5675075 (дата обращения: 02.10.2021).https://ieeexplore.ieee.org/document/5675075 (дата обращения: 02.10.2021).Zamani H., Nadimi-Shahraki M. H. Feature Selection Based on Whale Optimization Algorithm for Diseases Diagnosis // International Journal of Computer Science and Information Security. 2016. Vol. 14. Р. 1243‒1247.Zamani H., Nadimi-Shahraki M. H. Feature Selection Based on Whale Optimization Algorithm for Diseases Diagnosis // International Journal of Computer Science and Information Security. 2016. Vol. 14. Р. 1243‒1247.

The authors declare that there are no conflicts of interest present.