Data and fuzzy partition-based importance measure and its application in comprehensive evaluation

Comprehensive evaluation is an important part of modern decision making for practical problems such as resource management and complex system optimization. However, as the evaluation indices are always interactive and correlated, it is very difficult to determine the weight of index system. To solve this problem, the paper first proposes the interval neighborhood based on decision information system with numerical attributes. Then, fuzzy partition coverings in terms of the neighborhood are generated. By taking the interval neighborhood as the similarity description strategy, the positive region in the covering rough sets as knowledge carrier of the information system. It shows a metric to measure the hidden importance of the attributes. The evaluation model is constructed by using Choquet integral as an aggregation operator. The illustrated example validates the effectiveness of the method.