Variable stage self-correcting grey clustering model

Abstract

The grey clustering model based on the possibility function is frequently used in system evaluation, but the setting of the possibility function is frequently subjective, which results in the weak credibility of the clustering results. In addition, the "maximum criterion" decision paradox of the model has fuzzy boundaries, making the clustering process impossible to program. Based on this, the possibility function is represented as a matrix in the paper, and the possibility function is calculated by an optimization model to guarantee the objectivity of the clustering results. A variable stage self-correcting grey clustering model is built by using the kernel weight transformation as the primary tool, "entropy subtraction" as the signal of the decision paradox, and "entropy non-subtraction" as the model's termination signal. The model's Matlab source code is provided to reduce the complexity of the model. The illustration and comparative analysis demonstrate the model's logic and efficacy. Using the model for the evaluation of the meteorological drought grade in Henan Province can offer a fresh perspective for meteorological departments to make choices and assessments.