Comparativo entre Random Forest e Redes Neurais Artificiais na previsão de séries temporais sazonais e não estacionárias / Comparative between Random Forest and Artificial Neural Networks in forecasting seasonal and non-stationary time series
DOI:
https://doi.org/10.34117/bjdv8n1-235Keywords:
Random Forest, Séries Temporais, Redes Neurais ArtificiaisAbstract
Neste estudo, fez-se um comparativo dos resultados das previsões de dois grupos de séries temporais: um deles formado por séries estacionárias não sazonais (eNs) e outro formado por séries sazonais e não estacionárias (sNe). As previsões, de ambos os grupos, foram obtidas por meio dos métodos Random Forest (RF) e Redes Neurais artificiais (RNA) Feedforward. Além disso, comparou-se os resultados obtidos a partir de três estratégias de previsões multipassos à frente, a saber: Direta, Recursiva e MIMO (múltiplas entradas e múltiplas saídas). Foram utilizadas sessenta e duas séries temporais divididas entre os grupos eNs e sNe. As Redes Neurais Artificiais mostraram-se mais eficientes que o Random Forest nas previsões de ambos os grupos de séries temporais, sendo constatada uma maior diferença no grupo formado pelas séries sazonais e não estacionárias. Neste grupo, as previsões a partir da estratégia MIMO apresentaram RMSE (raiz do erro quadrático médio) de 386,51 com o uso do RF e 27,62 por meio da RNA.
References
AN, N. H.; ANH, D.T. Comparison of Strategies For Multi-Step-Ahead Prediction Of Time Series Using Neural Network. Anais... Ho Chi Minh City, Vietnam: International Conference On Advanced Computing And Applications, 2015.
BERGMEIR, C.; BENITEZ, J. M. Neural Networks in R Using the Stuttgart Neural Network Simulator: RSNNS. Journal of Statistical Software, 46(7), 1-26. URL http://www.jstatsoft.org/v46/i07/, 2012
BONTEMPI, G. Long Term Time Series Prediction With Multi-Input Multi-Output Local Learning. Anais... Helsinki, Finland: European Symposium on Time Series Prediction, 2008.
BREIMAN, L. Bagging predictors. Machine Learning, v.24, p.123-140, 1996.
BREIMAN, L. Random forests. Machine Learning, v.45, p.5-32, 2001.
CHEN, W. et al. Performance evaluation of the GIS-based data mining techniques of best-first decision tree, random forest, and naïve Bayes tree for landslide susceptibility modeling. Science of the Environment, v.644, p.1006-1018, 2018.
FACELI, K. et al. Inteligência Artificial: uma abordagem de aprendizado de máquina. Rio de Janeiro: LTC, 2011.
HAYKIN, S., Redes Neurais Princípios e Aplicações, 2. Ed. Porto Alegre: Bookman, 2001.
IBRAIM, A.; KASHEF, R.; CORRIGAN, L. Predicting market movement Direction for bitcoin: A comparison of time series modeling methods. Computers & Electrical Engineering, v.89, 2021.
JAMES, G. et al. An Introduction Statistical Learning: with applications in R. New York: Spinger, 2013.
KANG, Y.; LI, F. gratis: Generating Time Series with Diverse and Controllable Characteristics. R package version 0.2.0. https://CRAN.R-project.org/package=gratis, 2020.
KOEHRSEN, W. Random Forest Simple Explanation. Disponível em: https://medium.com/@williamkoehrsen/random-forest-simple-explanation-377895a60d2d. Acesso em: 24/10/2020.
LI, L. et al. Prediction of plant transpiration from environmental parameters and relative leaf area index using the random forest regression algorithm. Journal of Cleaner Production, v.261, 2020.
LI, R. et al. Estimating historical SO2 level across the whole China during 1973 e 2014 using random forest model. Chemosphere, v.247, 2020.
LIAW, A.; WIENER, M. Classification and Regression by randomForest. R News 2(3), 18--22, 2002.
LIU, D. et al. Random forest regression evaluation model of regional flood disaster resilience based on the whale optimization algorithm. Journal of Cleaner Production, v.250, 2020.
OLLECH, D. seastests: Seasonality Tests. R package version 0.14.2. https://CRAN.R-project.org/package=seastests, 2019.
R CORE TEAM. R: A language and environment for statistical computing. R Foundation for Statistical Computing, 2020. Vienna, Austria. Diponível em: https://www.R-project.org/. Acesso em: 23/10/2020.
SHIRI, J. Improving the performance of the mass transfer-based reference evapotranspiration estimation approaches through a coupled waveletrandom forest methodology. Journal of Hydrology, v.561, p.737-750, 2018.
SOUZA, R. C.; CAMARGO, M.E. Análise e Previsão de Séries Temporais: os modelos ARIMA. Ijuí: SEDIGRAF, 2004.
STEPCHENKO, A. et al. Nonlinear, Non-stationary and Seasonal Time Series Forecasting Using Different Methods Coupled with Data Preprocessing. Procedia Computer Science, v. 104, p. 578-585, 2017.
TAIEB, S. B. et al. A Review and Comparison of Strategies For Multi-step Ahead Time Series Forecating Based on the NN5 Forecasting Competition. Expert Systems With Applications, v.39, p.7067-7083, 2012.
TRAPLETTI, A; HORNIK, K. tseries: Time Series Analysis and Computational Finance. R package version 0.10-47, 2019.
TYRALIS, H.; PAPACHARALAMPOUS, G. Variable Selectionin Time Series Forecasting Using Random Forests. Algorithms, v.10, p.114, 2017.
VASSALLO, D. et al. Analysis of Random Forest Modeling Strategies for Multi-Step Wind Speed Forecasting. Energies, v. 13, 2020.
WONG, C. S.; LI, W. K. On a mixture autoregressive models. Journal of the Royal Statistical Society: Series B (Statistical Methodology), v.62, n.1, p. 95–115, 2000.
YANLAI, Z. et al. Multi-Output Support Vector Machine For Regional Multi-Step-Ahead PM2.5 Forecasting. Science of the Total Environment, v.651, p. 230-240, 2019.
ZHAO, W. et al. A spatial downscaling approach for the SMAP passive surface soil moisture product
