The existing three-way decisions mainly focus on various types of complete information systems or incomplete single-type information systems, while the data in the real application fields often exhibit incompleteness and complexity. Therefor this paper proposed the three-way decisions model and rule acquisition method for incomplete composite decision system based on a new complete neighborhood tolerance relation. First, this algorithm designed a new complete neighborhood tolerance classes under the incomplete mixed data, which could substitute for equivalence classes to calculate the conditional probabilities in the classical three-way decisions model. Then, it improved the concept of the loss function interval to obtain different thresholds of each object under the optimistic, compromising and pessimistic decisions and constructed there three-way decisions models and rule acquisition method under three decision-making risks for the incomplete composite decision system. Finally, this paper analyzed the validity and interpretable of the proposed algorithm in detail by a series of theoretical analysis and a medical diagnosis example. The experimental results based on different datasets show that the proposed model outperforms the state-of-the-art models in terms of classification process and expands the theory and application of three-way decisions and knowledge discovery.

[2020-05-05] Printed Article /doc/2018.10.0806/v2