A Multi-attribute Fuzzy Fluctuation Time Series Model Based on Neutrosophic Soft Sets and Information Entropy

Most existing forecasting models abstract logical rules that are based on historical discrete states in time series, and inconsistencies between these discrete states are rarely described quantitatively. In this paper, a multi-attribute fuzzy fluctuation time series-forecasting model based on neutrosophic soft sets (NSSs) and information entropy is proposed, which describes the complex changes of the logical rule training stage from the two characteristics of the state and volatility. The innovation and advantages of the model are mainly as follows: (1) The NSSs which have multi-attribute mapping and multi-dimensional expression functions can depict the complex state of multiple attributes in a specific period of time and thus characterise the state of the stock market clearly. (2) Using information entropy to quantify the degree of inconsistency of stock market fluctuations at a certain time which reflect the characteristics of volatility in the stock market effectively. (3) The similarity measure is used to find the optimal rule from the dimensions of state and volatility. To verify the validity of this model, this paper takes the Taiwan Stock Exchange Capitalization Weighted Stock Index (TAIEX) as an example. Experiments show that the model has a stable prediction performance for different data sets. Meanwhile, the prediction error is compared with other methods, which proves that the model has better prediction accuracy and versatility.