A Novel Fuzzy Time Series Forecasting Model Based on the Hybrid Wolf Pack Algorithm and Ordered Weighted Averaging Aggre
- PDF / 1,026,624 Bytes
- 19 Pages / 595.276 x 790.866 pts Page_size
- 36 Downloads / 177 Views
A Novel Fuzzy Time Series Forecasting Model Based on the Hybrid Wolf Pack Algorithm and Ordered Weighted Averaging Aggregation Operator Sidong Xian1
•
Tangjin Li2 • Yue Cheng1
Received: 18 July 2019 / Revised: 18 April 2020 / Accepted: 15 June 2020 Ó Taiwan Fuzzy Systems Association 2020
Abstract The fuzzy time series has received extensive attention since it was proposed and it has been widely used in various practical applications. This study proposes a new fuzzy time series forecasting model which considers a hybrid wolf pack algorithm (HWPA) and an ordered weighted averaging (OWA) aggregation operator for fuzzy time series. The HWPA is adopted to obtain a suitable partition of the universe of discourse to promote the forecasting performance. Furthermore, the improved OWA aggregation method is applied to make the aggregation of historical information more practical. To overcome the deficiency of slow convergence speed and easy to entrap into the local extremum of the wolf pack algorithm (WPA), the chemotactic behavior and elimination–dispersal behavior of bacterial foraging optimization (BFO) are employed to optimize the scouting behavior of WPA. The actual enrollments data of the University of Alabama and Taiwan Futures Exchange (TAIFEX) are utilized as the benchmark data and the computational results of both training and testing phases all indicate that the new forecasting model outperforms other existing models. The robustness of the proposed model is also tested and the robust results can be obtained when the historical data are inaccurate.
& Sidong Xian [email protected] 1
Key Laboratory of Intelligent Analysis and Decision on Complex Systems, Chongqing University of Posts and Telecommunications, Chongqing 400065, China
2
School of automation, Chongqing University of Posts and Telecommunications, Chongqing 400065, China
Keywords Fuzzy time series Wolf pack algorithm Bacterial foraging optimization Ordered weighted averaging aggregation operator Hybrid wolf pack algorithm TAIFEX
1 Introduction The theory of fuzzy sets was firstly proposed by Zadeh [1]. Song and Chissom [2–4] introduced the fuzzy set theory into the time series model successfully and proposed the fuzzy time series model. Compared with the time series model, the fuzzy time series prediction model has better prediction performance when dealing with small data sequences with incomplete, uncertain, and ambiguous correlation characteristics. Therefore, it has been received extensive attention from researchers. The research shows that the division of the universe of discourse is one of the key factors that influence the prediction accuracy of the fuzzy time series model. In the early study stage of the fuzzy time series model, Song and Chissom [2–4] and Chen [5] mainly used the equal division method to partition the fuzzy interval. This method is straightforward and easy to interpret. However, the defect of this method is that it does not consider the specific distribution of data, so the prediction accuracy of the model is not satis
Data Loading...
