A new ensemble intuitionistic fuzzy-deep forecasting model: Consolidation of the IFRFs-bENR with LSTM

CAĞCAĞ YOLCU Ö., YOLCU U.

Information Sciences, cilt.679, 2024 (SCI-Expanded) identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 679
  • Basım Tarihi: 2024
  • Doi Numarası: 10.1016/j.ins.2024.121007
  • Dergi Adı: Information Sciences
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, Aerospace Database, Applied Science & Technology Source, Business Source Elite, Business Source Premier, Communication Abstracts, Compendex, Computer & Applied Sciences, INSPEC, Library, Information Science & Technology Abstracts (LISTA), Metadex, MLA - Modern Language Association Database, zbMATH, Civil Engineering Abstracts
  • Anahtar Kelimeler: Elastic-net regularization, Forecasting, Genetic algorithm, Intuitionistic fuzzy regression functions, Linear and nonlinear relations, LSTM
  • Marmara Üniversitesi Adresli: Evet

Özet

Among forecasting model families, the intuitionistic fuzzy-based forecasting model stands out due to its comprehensive approach to uncertainty, considering possible degrees of hesitation. This study offers a forecasting model that consolidates intuitionistic fuzzy regression functions based on elastic net regularization (IFRFs-bENR) with LSTM. The proposed consolidated model, unlike existing models, is capable of modelling both linear and nonlinear structures that coexist between inputs and outputs. Another noteworthy aspect of the consolidated forecasting model is its method of determining model hyperparameters through a systematic optimization process using GA, in contrast to the trial-and-error approach prevalent in most literature studies. The validity and consistency of the model were assessed by running the model 50 times with the optimal hyperparameter values obtained for the consolidated model. And thus, the experimental probability distributions of the forecasts were also obtained. The proposed consolidated model also outperforms its peers in this aspect. The consolidated forecasting model was applied to different sets of time series, including TAIEX, DJI, SSEC, and IstEX. The findings indicate that the proposed consolidated model produces more accurate forecasts compared to various selected benchmark models. All abbreviations used in the article are defined in Supplementary Table 1 under the List of Abbreviations.