Neutrosophic Soft Sets Forecasting Model for Multi-Attribute Time Series

Traditional time series forecasting models mainly assume a clear and definite functional relationship between the historical values and the current/future values of a dataset. In this paper, we extended the current model by generating multi-attribute forecasting rules based on the consideration of combining multiple related variables. In this model, neutrosophic soft sets (NSSs) are employed to represent historical statues of several closely related attributes in stock market, such as volumes, stock market index, and daily amplitudes. Specifically speaking, the status of up, equal, and down in historical stock index can be represented by truth, indeterminacy, and false, respectively, by neutrosophic sets (NSs) and NSSs can build mappings of different related attributes to NSs. The advantages of the proposed model are: 1) using NSSs to enclose different historical characteristics in time series to preserve inherent complexity of a dataset with mapping adequate features and 2) with the existing researches of NSSs, it is efficient with using Euclidean distance to find the optimal rules, and thus, the model can avoid incomplete of rules due to the limited sample dataset. To evaluate the performance of the model, we explored the closing prices of the Taiwan Stock Exchange Capitalization Weighted Stock Index (TAIEX) as the major parameter that we forecast and the stock amplitudes and volumes as the other factors to facilitate the predicting of the TAIEX. To show the universality of the model, we applied the proposed model to forecast some other influential indexes as well.