kamniroИсследования водных биологических ресурсов Камчатки и северо-западной части Тихого океанаThe researches of the aquatic biological resources of Kamchatka and the North-West Part of the Pacific Ocean2072-82122782-6236«КамчатНИРО»10.15853/2072-8212.2020.59.76-96kamniro-154Research ArticleОРИГИНАЛЬНЫЕ НАУЧНЫЕ СТАТЬИFULL ARTICLESИспользование метода Random Forest в целях прогнозирования подходов горбуши северо-востока КамчаткиUsing the decisions of the Random Forest algorithm for the purposes of forecasting pink salmon runs on north-eastern KamchatkaФельдманМ. Г.FeldmanM. H.

Вед. н. с.

683000 Петропавловск-Камчатский, Набережная, 18

Тел.: 8 (4152) 41-27-01

Leading Scientist

683000 Petropavlovsk-Kamchatsky, Naberezhnaya Str., 18

Tel.: +7 (4152) 41-27-01

feldman.m.g@kamniro.ruКамчатский филиал Всероссийского научно-исследовательского института рыбного хозяйства и океанографии («КамчатНИРО»)РоссияKamchatka Branch of Russian Research Institute of Fisheries and Oceanography (“KamchatNIRO”)Russian Federation2020180920200597696Copyright © Фельдман М.Г., 20202020Фельдман М.Г.Feldman M.H.Данная работа распространяется под лицензией Creative Commons Attribution 4.0.This work is licensed under a Creative Commons Attribution 4.0 License.https://kamniro.elpub.ru/jour/article/view/154

Для прогнозирования подходов камчатской горбуши используется мощный современный метод машинного обучения Random Forest — случайный лес деревьев решений. В качестве предикторов используются помесячные данные климатических индексов. В работе применяется итеративный способ отбора наиболее важных факторов. Выбор лучшей модели осуществлен по наименьшей ошибке на тестовых данных. Алгоритм применяемого метода оформлен на языке R.

Forecasting of pink salmon runs in Kamchatka uses modern powerful method of machine learning Random Forest (random forest of decision trees). Monthly data of climate indices are used as predictors. Forecasting applies iterative way of selection of the most important factors. Decision about the best model is based on the least error on test data. The algorithm of the method is written in R language.

прогнозированиегорбушаклассификациямоделированиеслучайный лесдерево решенийязык rанализ данныхforcastingpink salmonclassificationmodelingrandom forestcode rdata miningReferencesБугаев А.В., Тепнин О.Б., Радченко В.И. 2018. Климатическая изменчивость и продуктивность тихоокеанских лососей Дальнего Востока России // Исслед. водн. биол. ресурсов Камчатки и сев.-зап. части Тихого океана. Вып. 49. С. 5–50.Bugaev A.V., Tepnin O.B., Radchenko V.I. Climate variability and Pacific salmon productivity in Russian Far East. The researches of the aquatic biological resources of Kamchatka and of the north–west part of the Pacific Ocean, 2018, vol. 49, pp. 5–50. (In Russian with English abstract)Джеймс Г., Уиттон Д., Хасти Т., Тибширани Р. 2016. Введение в статистическое обучение с примерами на языке R. Пер. С.Э. Мастицкого. М.: ДМК Пресс, 450 с.James G., Whitton D., Hastie T., Tibshirani R. An Introduction to Statistical Learning with Applications. Moscow: DMK Press, 2016, 450 p.Карпенко В.И. 1998. Ранний морской период жизни тихоокенских лососей: Монография. М.: ВНИРО, 165 с.Karpenko V.I. Ranniy morskoy period zhizni tikhookenskikh lososey [Early marine period of life of Pacific Ocean salmon: monograph]. Moscow: VNIRO, 1998, 165 p.Кляшторин Л.Б., Любушин А.А. 2005. Циклические изменения климата и рыбопродуктивности. М.: ВНИРО, 235 с.Klyashtorin L.B., Lyubushin A.A. Tsiklicheskiye izmeneniya klimata i ryboproduktivnosti [Cyclical changes in climate and fish productivity]. Moscow: VNIRO, 2005, 235 p.Маркевич Н.Б., Виленская Н.И. 1998. Влияние сроков нереста и термического режима на выживаемость и рост молоди горбуши Оncorhinchus gorbusha на ключевых и русловых нерестилищах Западной Камчатки // Исслед. биологии и динамики численности промысловых рыб Камчатского шельфа. Вып. I. Ч. 1. С. 85–104.Markevich N.B., Vilenskaya N.I. Effects of the time and thermal regime of spawning on the survival and growth of juvenile pink salmon Oncorhynchus gorbusha in the spring and river main body spawning grounds of Western Kamchatka. The researches of the aquatic biological resources of Kamchatka and of the north-west part of the Pacific Ocean, 2018, vol. 49, pp. 5–50. (In Russian)Радченко В. 2017. Открытый курс машинного обучения. Тема 5. Композиции: бэггинг, случайный лес. Электронный блог компании Open Data Science, адрес доступа: https://habr.com/ru/company/ods/blog/324402/.Radchenko V. Machine Learning Open Course. Topic 5. Compositions: bagging, random forest. 2017, Electronic blog of the Open Data Science company, access address: https://habr.com/ru/company/ods/blog/324402/Фельдман М.Г., Шевляков Е.А. 2015. Выживаемость камчатской горбуши как результат совокупного воздействия плотностной регуляции и внешних факторов среды // Изв. ТИНРО. Т. 182. С. 88–114. Фельдман М.Г., Шевляков Е.А., Артюхина Н.Б. 2018. Оценка ориентиров пропуска производителей тихоокеанских лососей в бассейнах рек Северо-Восточной Камчатки // Исслед. водн. биол. ресурсов Камчатки и сев.-зап. части Тихого океана. Вып. 51. С. 5–26.Feldman M.G., Shevlyakov E.A. Survival of Kamchatka pink salmon as a result of combined effect of density and environmental regulating factors. Izvestia TINRO, 2015, vol. 182, pp. 88–114. (In Russian) Feldman M.G., Shevlyakov E.A., Artukhina N.B. Evaluation of teh Pacific salmon spawning escapement parameters for the river basins of North-East Kamchatka. The researches of the aquatic biological resources of Kamchatka and of the north-west part of the Pacific Ocean, 2018, vol. 51, pp. 5–26. (In Russian) Shitikov V.K., Mastitsky S.E. Classification, regression and R-using algorithms of data mining. Electronic book, 2017, access address: https://github.com/ranalytics/data-miningШитиков В.К., Мастицкий С.Э. 2017. Классификация, регрессия, алгоритмы Data Mining с использованием R. Электронная книга, адрес доступа: https://github.com/ranalytics/data-mining.Shuntov V.P., Temnykh O.S. Main results of the TINROCenter research of marine period of Pacific salmon. Izvestia TINRO, 2005, vol. 141, pp. 30–55. (In Russian) Shuntov V.P., Temnykh O.S. Pacific salmon in marine and ocean ecosystems: monograph. Vladivostok: TINRO-Center, 2011, vol. 2, 473 p.Шунтов В.П., Темных О.С. 2005. Основные результаты изучения морского периода жизни тихоокеанских лососей в ТИНРО-Центре // Изв. ТИНРО. Т. 141. С. 30–55.Surowiecki J. The Wisdom of Crowds: Why the Many are Smarter than the the Few and How Collective Wisdom Shapes Business, Economies, Societies and Nations. New York: Anchor Books, 2005. 304 p. (Translated into Russian)Шунтов В.П., Темных О.С. 2011. Тихоокеанские лососи в морских и океанических экосистемах: Монография. Т. 2. Владивосток: ТИНРО-Центр, 473 с.Breiman L. Bagging Predictors. Machine Learning, 1996a, № 24, pp. 123–140.Шуровьески Дж. 2007. Мудрость толпы. Почему вместе мы умнее, чем поодиночке, и как коллективный разум формирует бизнес, экономику, общество и государство. Пер. с англ. М.: ООО «И.Д. Вильяме», 304 с.Breiman L. Out-of-bag estimation. Technical report, Dept. of Statistics, Univ. of Calif., Berkeley, 1996b. Electronic source, access address: https://www.stat.berkeley.edu/~breiman/OOBestimation.pdf.Breiman L. 1996a. Bagging Predictors // Machine Learning: journal. Vol. 24, no. 2. P. 123–140.Breiman L., Friedman J.H., Olshen R.A., Stone C.J. Classification and regression trees. Wadsworth International Group, Belmont CA, 1984, 368 p.Breiman L. 1996b. Out-of-bag estimation. Technical report, Dept. of Statistics, Univ. of Calif., Berkeley. Электронный источник, адрес доступа: https:// www.stat.berkeley.edu/~breiman/OOBestimation.pdf. Breiman L. 2001. Random Forests // Machine Learning: journal. Vol. 45, no. 1. P. 5–32.Delgado F.M., Cernadas E., Barro S., Amorim D. Do we need hundreds of сlassifiers to solve real world classification problems? J. of Machine Learning Research, 2014, № 15, pp. 3133–3181.Breiman L., Friedman J.H., Olshen R.A., Stone C.J. 1984. Classification and regression trees. Wadsworth International Group, Belmont CA, 368 p.Efron B. Bootstrap Methods: Another Look at the Jackknife. Annals of Statistics, 1979, vol. 7. № 1, pp. 1–26. Galton F. Vox populi. Nature, 1907, № 75, pp. 450– 451.Delgado F.M., Cernadas E., Barro S., Amorim D. 2014. Do we need hundreds of сlassifiers to solve real world classification problems? // J. of Machine Learning Research, № 15. P. 3133–3181.Haeseker S., Dorner B., Peterman R., Su Z. An improved sibling model for forecasting Chum Salmon and Sockeye Salmon abundance. North American Journal of Fisheries Management, 2007, № 27, pp. 634–642. Hare S.R. Low frequency climate variability and salmon production. Ph.D. Dissertation. Univ. of Washington, Seattle, WA, 1996, 306 p.Efron B. 1979. Bootstrap Methods: Another Look at the Jackknife. Annals of Statistics, Vol. 7. № 1. P. 1–26. Galton F. 1907. Vox populi // Nature, № 75. P. 450– 451.Hare S.R., Francis R.C. In Climate Change and Northern Fish Populations. Can. Spec. Publ. Fish. Aquat. Sci., 1995, vol. 121, pp. 357–372.Haeseker S., Dorner B., Peterman R., Su Z. 2007. An improved sibling model for forecasting Chum Salmon and Sockeye Salmon abundance // North American Journal of Fisheries Management. № 27. P. 634–642. Hare S.R. 1996. Low frequency climate variability and salmon production. Ph.D. Dissertation. Univ. of Washington, Seattle, WA, 306 p.Ho T.K. Random Decision Forests. Proceedings of the 3rd International Conference on Document Analysis and Recognition. Montreal, QC, 14–16 August, 1995, pp. 278–282.Hare S.R., Francis R.C. 1995. Climate change and salmon production in the Northeast Pacific Ocean // In Climate Change and Northern Fish Populations, ed. by R.J. Beamish, Can. Spec. Publ. Fish. Aquat. Sci. Vol. 121. P. 357–372.Kleinberg E. Stochastic Discrimination. Annals of Mathematics and Artificial Intelligence, 1990, vol. 1 (1–4), pp. 207–239.Ho T.K. 1995. Random Decision Forests. Proceedings of the 3rd International Conference on Document Analysis and Recognition. Montreal, QC: 14–16 August. P. 278–282.Kleinberg E. An Overtraining-Resistant Stochastic Modeling Method for Pattern Recognition. Annals of Statistics, 1996, vol. 24 (6), pp. 2319–2349.Kleinberg E. 1990. Stochastic Discrimination // Annals of Mathematics and Artificial Intelligence, Vol. 1 (1–4). P. 207–239.Kursa M. Boruta for those in a hurry. 2020. Electronic source, access address: https://cran.r-project.org/web/packages/Boruta/vignettes/inahurry.pdf.Kleinberg E. 1996. An Overtraining-Resistant Stochastic Modeling Method for Pattern Recognition // Annals of Statistics. Vol. 24 (6). P. 2319–2349.Kursa M., Rudnicki W. Feature Selection with the Boruta Package. J. of Statistical Software, 2010, vol. 36 (11), pp. 2–12.Kursa M. 2020. Boruta for those in a hurry. Электронная статья, адрес доступа: https://cran.r-project.org/web/packages/Boruta/vignettes/inahurry.pdf.Linkin M.E., Nigam S. The North Pacific Oscillation – West Pacific Teleconnection Pattern: Mature-Phase Structure and Winter Impacts. J. Climate, 2008, vol. 21, № 9, pp. 1979–1997.Kursa M., Rudnicki W. 2010. Feature Selection with the Boruta Package // J. of Statistical Software. Vol. 36 (11). P. 2–12.Mantua N., Hare S., Zhang Y., Wallace J., Francis R.A. Pacific interdecadal climate oscillation with impacts on salmon production. Bull. Amer. Meteor. Soc., 1997, № 78, pp. 1069–1079.Linkin M.E., Nigam S. 2008. The North Pacific Oscillation – West Pacific Teleconnection Pattern: Mature-Phase Structure and Winter Impacts // J. Climate. Vol. 21. № 9. P. 1979–1997.Mantua N.J., Hare S.R. The Pacific Decadal Oscillation. J. of Oceanography, 2002, vol. 58, pp. 35–44. Paluszyńska A. Understanding random forests with randomForestExplainer. 2021. Electronic source, access address: https://cran.rstudio.com/web/packages/randomForestExplainer/vignettes/randomForestExplainer.html.Mantua N., Hare S., Zhang Y., Wallace J., Francis R. 1997. A Pacific interdecadal climate oscillation with impacts on salmon production // Bull. Amer. Meteor. Soc., № 78. P. 1069–1079.Peterman R.M. Model of salmon age structure and its use in preseason forecasting and studies of marine survival. Canadian Journal of Fisheries and Aquatic Sciences, 1982, № 39, pp. 1444–1452.Mantua N.J., Hare S.R. 2002. The Pacific Decadal Oscillation // J. of Oceanography. Vol. 58. P. 35– 44.Quinlan J.R. Induction of Decision Trees. Machine Learning. Kluwer Academic Publishers, 1986, № 1, pp. 81–106.Paluszyńska A. Understanding random forests with randomForestExplainer. Электронная статья, адрес доступа: https://cran.rstudio.com/web/packages/randomForestExplainer/vignettes/randomForestExplainer.html.Ricker W.E. Stock and Recruitment. J. of the Fisheries Research Board of Canada, 1954, vol. 11, № 5, pp. 559–623.Peterman R.M. 1982. Model of salmon age structure and its use in preseason forecasting and studies of marine survival // Canadian Journal of Fisheries and Aquatic Sciences. № 39. P. 1444–1452.Thompson D., Wallace J. The Arctic Oscillation signature in the wintertime geopotential height and temperature fields. Geophys. Res. Lett., 1998, vol. 25, № 9, pp. 1297–1300.Quinlan J.R. 1986. Induction of Decision Trees // Machine Learning. Kluwer Academic Publishers. № 1. P. 81–106.Quinlan J.R. 1986. Induction of Decision Trees // Machine Learning. Kluwer Academic Publishers. № 1. P. 81–106.Ricker W.E. 1954. Stock and Recruitment // J. of the Fisheries Research Board of Canada. Vol. 11. № 5. P. 559–623.Ricker W.E. 1954. Stock and Recruitment // J. of the Fisheries Research Board of Canada. Vol. 11. № 5. P. 559–623.Thompson D., Wallace J. 1998. The Arctic Oscillation signature in the wintertime geopotential height and temperature fields. Geophys. Res. Lett., Vol. 25. № 9. P. 1297–1300.Thompson D., Wallace J. 1998. The Arctic Oscillation signature in the wintertime geopotential height and temperature fields. Geophys. Res. Lett., Vol. 25. № 9. P. 1297–1300.

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