Finally, apply these C646 weights to the indicator system proposed in Section 2.1, we finish the construction of safety evaluation indicator system.
3. Improved grey variable weight clustering evaluation model
Evaluation indicator systems in most existing studies have one thing in common: the regions of evaluation levels have clear boundaries. This means two different evaluations will be viewed as belonging to the same safety level if their indicator values fall into the same region. In reality, results from these methods are often regarded as incomplete and biased (Hartford, Carey, & Mendonca, 2007). Thus static deterministic evaluation methods should be improved.
Su et al. (2010) provided a fuzzy judgment method to integrate multi-factor values. It helps to obtain a more comprehensive evaluation but its fuzzy judgment matrix (the membership degrees of factors to judgment sets) is assessed subjectively. So, nares paper provides a new grey variable weight clustering method to improve evaluation where boundaries of risk levels are vague.