ABSTRACT
Forecasting is one of the most important tasks in temporal data mining. Actually, most of forecasting applications require performing multistep forecasts. Thus, what is the appropriate multistep forecasting strategy to use with each model? To answer this question, this study evaluates the impact of multistep forecasting strategies on the performance and the stability of recurrent neural networks models (SRN, LSTM and GRU) for short term and long term horizons. This comparison is based on well-defined benchmarking univariate time series (deterministic, stochastic and chaotic), whose properties are known and challenged their modeling.
The results of this study proved that direct and hybrid strategies are computationally expensive than multiple outputs and recursive strategies. Further, we had come out with the following guidelines: For SRN and LSTM models, multiple outputs strategy is suggested for short term forecasting for all types of time series. While for long term forecasting, multiple outputs is suggested for stochastic time series, however, direct strategy for deterministic data and hybrid strategy for chaotic data. On the other hand, for GRU model, multiple outputs strategy is suggested for short and long horizons for all data types, except for the case of long term forecasts with chaotic data, where direct or hybrid are preferred. Thereupon, Switching-based forecasting system is suggested as a relevant alternative to build up accurate predictions for each model.
- Mamoulis N., 2009. Temporal Data Mining. In: LIU L., ÖZSU M.T. (eds) Encyclopedia of Database Systems. Springer, Boston, MA. DOI: https://doi.org/10.1007/978-0-387-39940-9_393.Google Scholar
- Fu, T.C., 2011. A review on time series data mining. Engineering Applications of Artificial Intelligence 24, 164--181. DOI: 10.1016/j.engappai.2010.09.007.Google ScholarDigital Library
- Lin, W., Orgun, M. A., and Williams, G. J., 2002. An Overview of temporal data mining. In S. J. Simoff, G. J. Williams, & M. Hegland (Eds.), Proceedings, Australasian Data Mining Workshop, ADM02 (pp. 83--89). Sydney: University of Technology, Sydney.Google Scholar
- Yang, Y., 2017.Temporal Data Mining. Temporal Data Mining Via Unsupervised Ensemble Learning. 9--18. DOI:10.1016/b978-0-12-811654-8.00002-6.Google Scholar
- Box, G.E.P., Jenkins, G.M., Reinsel, G.C., and Ljung, G.M., 2015. Time Series Analysis: Forecasting and Control. Wiley Series in Probability and Statistics, 5, Wiley, New Jersey, USA. ISBN: 978-1-118-67502-1.Google Scholar
- Parmezan, A.R.S., Vinicius M.A.S. and Batista, G.E.A.P.A., 2019. Evaluation of statistical and machine learning models for time series prediction: Identifying the state-of-the-art and the best conditions for the use of each model. Information Sciences 484, 302--337. DOI: 10.1016/j.ins.2019.01.076.Google ScholarDigital Library
- Ahmed, N.K., Atiya, A.F., Gayar, N.E., and El-Shishiny, H., 2010. An empirical comparison of machine learning models for time series forecasting, Econ. Rev. 29 (5--6), 594--621. DOI: 10.1080/07474938.2010.481556.Google Scholar
- Ristanoski, G., Liu, W., and Bailey, J., 2013. A time-dependent enhanced support vector machine for time series regression. In the International Conference on Knowledge Discovery and Data Mining, ACM, Chicago, USA, 2013. 946--954. DOI: 10.1145/2487575.2487655.Google ScholarDigital Library
- Parmezan, A.R.S., and Batista, G.E.A.P.A., 2015. A study of the use of complexity measures in the similarity search process adopted by kNN algorithm for time series prediction. In the International Conference on Machine Learning and Applications, IEEE, Miami, USA. 45--51. DOI: 10.1109/ICMLA.2015.217.Google Scholar
- Khaldi, R., El Afia A., and Chiheb R., 2019. Forecasting of BTC volatility: comparative study between parametric and nonparametric models. Progress in Artificial Intelligence. https://doi.org/10.1007/s13748-019-00196-w.Google ScholarDigital Library
- Cortez, P., 2010. Sensitivity analysis for time lag selection to forecast seasonal time series using neural networks and support vector machines. In the International Joint Conference on Neural Networks, IEEE, Barcelona, Spain, 2010, pp.3694--3701. DOI: 10.1109/IJCNN.2010.5596890.Google ScholarCross Ref
- Zhang, X., Zhang, T., Young, A. A., and Li, X., 2014. Applications and comparisons of four time series models in epidemiological surveillance data, PLoS One 9 (2). DOI: 10.1371/journal.pone.0088075.Google Scholar
- Kandananond, K., 2012. A comparison of various forecasting methods for autocorrelated time series, Int. J. Eng. Bus. Manag.4, 1--6. DOI: 10.5772/51088.Google ScholarCross Ref
- Zhang X, Liu Y, Yang M, Zhang T, Young AA, and Li, X., 2013. Comparative Study of Four Time Series Methods in Forecasting Typhoid Fever Incidence in China. PLoS ONE 8(5): e63116. DOI:10.1371/journal.pone.0063116.Google ScholarCross Ref
- Divina, F., García Torres, M., Goméz Vela, F.A., and Vázquez Noguera, J.L. 2019. A Comparative Study of Time Series Forecasting Methods for Short Term Electric Energy Consumption Prediction in Smart Buildings. Energies 2019, 12, 1934.Google Scholar
- Hu, M., Li, W., Yan, K., Ji, Z., and Hu, H., 2019. Modern Machine Learning Techniques for Univariate Tunnel Settlement Forecasting: A Comparative Study. Mathematical Problems in Engineering. Vol. 2019, Article ID 7057612, 12 pages, 2019. DOI: 10.1155/2019/7057612.Google Scholar
- Cecati, C., Kolbusz, J., Różycki, P., Siano, P., and Wilamowski, B. M., 2015. A Novel RBF Training Algorithm for Short-Term Electric Load Forecasting and Comparative Studies. IEEE Transactions on Industrial Electronics. Volume: 62, Issue: 10, Oct. 2015. DOI 10.1109/TIE.2015.2424399.Google Scholar
- Renani E.T., Fathi, M., Elias, M., and Abd Rahim, N., 2016. Using data driven approach for wind power prediction: A comparative study. Energy Conversion and Management. Volume 118, 193--203. DOI: 10.1016/j.enconman.2016.03.078.Google ScholarCross Ref
- Gong, Y., Zhang, Y., Lan, S., and Wang, H., 2016. A Comparative Study of Artificial Neural Networks, Support Vector Machines and Adaptive Neuro Fuzzy Inference System for Forecasting Groundwater Levels near Lake Okeechobee, Florida. Water Resour Manage. Volume 30, Issue 1, 375--391. DOI 10.1007/s11269-015-1167-8.Google Scholar
- Choubin, B., Zehtabian, G., Azareh, A., Rafei-Sardooi, E., Sajedi-Hosseini, F., and Kisi, Ö, 2018. Precipitation forecasting using classification and regression trees (CART) model: a comparative study of different approaches. Environmental Earth Sciences. 77:314. DOI: 10.1007/s12665-018-7498-z.Google ScholarCross Ref
- Khaldi, R., El Afia, A., and Chiheb, R. 2019. Performance Prediction of Pharmaceutical Suppliers: A Comparative Study between DEA-ANFIS-PSO and DEA-ANFIS-GA, Int. J. Computer Applications in Technology. Vol. 60, No. 4, pp.317--325.Google ScholarDigital Library
- Khaldi, R., Chiheb R., and El Afia A., 2018. Feedforward and Recurrent Neural Networks for Time Series Forecasting: Comparative Study. In Proceedings of ACM LOPAL conference, Rabat, Morocco, May 2018 (LOPAL'18), 6 pages. DOI: 10.1145/3230905.3230946.Google ScholarDigital Library
- Khaldi, R., El Afia A. and Chiheb R., 2017. Artificial Neural Network Based Approach for Blood Demand Forecasting: Fez Transfusion Blood Center Case Study. 2nd BDCA conference. ACM. DOI: 10.1145/3090354.3090415.Google ScholarDigital Library
- Khaldi, R., Chiheb, R., El Afia A., Akaaboune, A., and Faizi, R., 2017. Prediction of Supplier Performance: A Novel DEA-ANFIS Based Approach. 2nd BDCA conference. ACM. DOI: 10.1145/3090354.3090416.Google ScholarDigital Library
- Khaldi, R., El Afia A., Chiheb R., and Faizi R., 2018. Forecasting of Bitcoin Daily Returns with EEMD-ELMAN based Model. In Proceedings of ACM LOPAL conference, Rabat, Morocco, May 2018 (LOPAL'18), 6 pages. DOI: 10.1145/3230905.3230948.Google ScholarDigital Library
- Khaldi, R., El Afia A., and Chiheb R., 2019. Forecasting of Weekly Patient Visits to Emergency Department: Real Case Study. Second International Conference on Intelligent Computing in Data Sciences (ICDS 2018). Procedia Computer Science. 148:532--541. DOI: 10.1016/j.procs.2019.01.026.Google ScholarDigital Library
- Sarhani, M., El Afia, A., and Faizi, R., 2017. Hybrid approach-based support vector machine for electric load forecasting incorporating feature selection. International Journal of Big Data Intelligence, 4(3), 141--148. DOI: 10.1504/IJBDI.2017.085520.Google ScholarCross Ref
- Sarhani, M. and El Afia, A, 2018. Forecasting Demand With Support Vector Regression Technique Incorporating Feature Selection in the Presence of Calendar Effect. In Contemporary Approaches and Strategies for Applied Logistics. 12:302--316. IGI Global. DOI: 10.4018/978-1-5225-5273-4.ch012.Google ScholarCross Ref
- Sarhani, M. and El Afia, A., 2017. Forecasting Demand with Support Vector Regression Technique Combined with X13-ARIMA-SEATS Method in the Presence of Calendar Effect. In Artificial Intelligence: Concepts, Methodologies, Tools, and Applications. 89:2146-2159. IGI Global. DOI: 10.4018/978-1-5225-1759-7.ch089.Google Scholar
- Sarhani, M. and El Afia, A., 2016. Feature selection and parameter optimization of support vector regression for electric load forecasting. In International Conference Electrical and Information Technologies (ICEIT) IEEE. 288--293. DOI: 10.1109/EITech.2016.7519608.Google Scholar
- Sarhani, M. and El Afia, A., 2014. Intelligent system based support vector regression for supply chain demand forecasting. In Second World Conference on Complex Systems. IEEE. 79--83. DOI: 10.1109/ICoCS.2014.7060941.Google Scholar
- Sarhani, M. and El Afia, A., 2015. Electric Load Forecasting Using Hybrid Machine Learning Model incorporating Feature selection. In Proceedings of the First International Conference on Big Data, Cloud and Applications. CEUR Workshop Proceedings.Google Scholar
- Bounabi, M., El Moutaouakil, K., and Satori, K., 2018. A Probabilistic Vector Representation and Neural Network for Text Classification. In Proceedings of the International Conference on Big Data, Cloud and Applications. Springer, Cham. 343--355. DOI: 10.1007/978-3-319-96292-4_27.Google Scholar
- Bounabi, M., El Moutaouakil, K., and Satori, K., 2017. A comparison of Text Classification methods Method of weighted terms selected by different Stemming Techniques. In Proceedings of the 2nd international Conference on Big Data, Cloud and Applications. ACM, 2017. p. 43. DOI: 10.1145/3090354.3090398.Google ScholarDigital Library
- Bounabi, M., El Moutaouakil, K., and Satori, K., 2019. A comparison of text classification methods using different stemming techniques. International Journal of Computer Applications in Technology (IJCAT). 60(4), 298--306. DOI: 10.1504/IJCAT.2019.101171.Google ScholarDigital Library
- Dahmouni, A., El Moutaouakil, K., and Satori, K., 2017. A Cloud Face Recognition System using A New Optimal Local Binary Pattern. In Proceedings of the 2nd international Conference on Big Data, Cloud and Applications, p. 39. ACM, 2017. DOI: 10.1145/3090354.3090394.Google ScholarDigital Library
- Ettaouil, M., Lazaar, M., and En-Naimani, Z., 2013. A hybrid ANN/HMM models for arabic speech recognition using optimal codebook. 8th International Conference on Intelligent Systems: Theories and Applications, SITA 2013. DOI: 10.1109/SITA.2013.6560806.Google Scholar
- En-Naimani, Z., Lazaar, M., and Ettaouil, M., 2016. Architecture Optimization Model for the Probabilistic Self-Organizing Maps and Speech Compression. International Journal of Computational Intelligence and Applications. 15 (2). DOI: 10.1142/S1469026816500073.Google Scholar
- Amrani, Y.A., Lazaar, M., and El Kadiri, K.E., 2018. A novel hybrid classification approach for sentiment analysis of text document. International Journal of Electrical and Computer Engineering. 8 (6). 4554--4567. DOI: 10.11591/ijece.v8i6.pp 4554--4567.Google Scholar
- Omara, H., Lazaar, M., and Tabii, Y., 2018. Effect of feature selection on gene expression datasets classification accuracy. International Journal of Electrical and Computer Engineering. 8 (5). 3194--3203. DOI: 10.11591/ijece.v8i5.pp.3194--3203.Google Scholar
- El Afia, A., Sarhani, M., and Aoun, O., 2017. Hidden markov model control of inertia weight adaptation for Particle swarm optimization. IFAC-PapersOnLine. Volume 50, Issue 1, July 2017, Pages 9997--10002. DOI: 10.1016/j.ifacol.2017.08.2030.Google Scholar
- Bouzbita, S., El Afia, A., Faizi, R., and Zbakh, M., 2016. Dynamic adaptation of the ACS-TSP local pheromone decay parameter based on the Hidden Markov Model. In the 2nd International Conference on Cloud Computing Technologies and Applications (CloudTech). DOI: 10.1109/CloudTech.2016.7847719.Google Scholar
- El Afia, A., Bouzbita, S., and Faizi, R., The effect of updating the local pheromone on ACS performance using fuzzy logic. International Journal of Electrical and Computer Engineering 7(4):2161. DOI: 10.11591/ijece.v7i4.pp2161-2168.Google Scholar
- Lalaoui, M., El Afia, A., and Chiheb, R., 2018. A self-tuned simulated annealing algorithm using hidden markov model. International Journal of Electrical and Computer Engineering. 8(1):291--298. DOI: 10.11591/ijece.v8i1. pp 291--298.Google Scholar
Index Terms
- Impact of Multistep Forecasting Strategies on Recurrent Neural Networks Performance for Short and Long Horizons
Recommendations
Feedforward and Recurrent Neural Networks for Time Series Forecasting: Comparative Study
LOPAL '18: Proceedings of the International Conference on Learning and Optimization Algorithms: Theory and ApplicationsThis study aims at examining and comparing the ability of ANNs variances, including MLP, RBFNN, ELMAN and JORDAN in forecasting monthly data of four different time series patterns. As well as to criticize the concept asserting that MLP is a "universal ...
Time series forecasting by recurrent product unit neural networks
Time series forecasting (TSF) consists on estimating models to predict future values based on previously observed values of time series, and it can be applied to solve many real-world problems. TSF has been traditionally tackled by considering ...
Improving Time Series' Forecast Errors by Using Recurrent Neural Networks
ICSCA '18: Proceedings of the 2018 7th International Conference on Software and Computer ApplicationsElman Neural Network (ENN) is considered one of the most powerful tool in solving various models. This paper suggests the use of ENN in a model free technique to solve time series models of any type. The objective of this paper is to compare between the ...
Comments