运用条件指标和方差分解比提出一种诊断设计阵复共线性的新方法——条件指标一方差分解比(CIVDP)法,该方法不但可以确定设计阵数据列之间复共线性关系的个数,而且可以确定每个复共线性关系存在于哪些数据列之间,并通过大量的模拟计算和实际应用验证该方法的有效性。
In this paper, a new method based on condition index and variance decomposition proportion (CIVDP) is proposed to diagnose the presence, number and composition of multicollinearities which exist among the columns of the design matrix in Gauss-Markov model for surveying adjustment. Some numerical examples are put forward which can prove the effectiveness and advantage of the new method.