A comparative study of nonlinear regression and autoregressive techniques in hybrid with particle swarm optimization for time-series forecasting. (15th March 2022)

• Record Type: Journal Article
• Authors: Kuranga, C.; Pillay, N.
• Is Part Of: Expert systems with applications. Volume 190(2022)
• Publication Date: 2022-03-15
• Subjects: Time-series forecasting -- Least-squares -- Nonlinear autoregressive -- Concept shifts -- Passive learning -- Quantum-inspired particle swarm optimization; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2021.116163
• Languages: English
• ISSNs: 0957-4174

Abstract

Highlights: QPSO-based technique induces an adaptive prediction model in a dynamic environment. QPSO-based technique hybridizes a QPSO with either a QR decomposition or a NARX. QPSO-based technique is evaluated on six time-series datasets. The performance of NARX-QPSO is competitive to several state-of-the-art techniques. Abstract: Usually, real-world time-series forecasting problems are dynamic. If such time-series are characterized by mere concept shifts, a passive approach to learning become ideal to continuously adapt the model parameters whenever new data patterns arrive to cope with uncertainty in the presence of change. This work hybridizes a quantum-inspired particle swarm optimization designed for dynamic environments, to cope with concept shifts, with either a least-squares approximation technique or nonlinear autoregressive model to forecast time-series. Also, this work evaluates experimentally and performs a comparative study on the performance of the proposed models. The obtained results show that the nonlinear autoregressive-based model outperformed the least-squares approximation-based model and the separate models that were implemented in the hybridization and also, several state-of-the-art models for the given datasets.