重尾数据的Coxian分布近似表示

摘要

研究了重尾数据的Coxian表示问题，建立了求解Coxian分布的相关参数的非线性规划数学模型．利用遗传算法求解该模型，并给出了相应的迭代算法．引入距离测度公式比较了不同阶次的Coxian分布的误差，找到了最优阶次的Coxian拟合．借助于Matlab软件模拟了一个数值，结果是令人满意的．%We studied a Coxian-representation problem on Heavy-tailed data and established the nonlinear programming model to analyze the related parameter of the Coxian probability distribution. Using geneticalgorithm and corresponding iterative steps, we can solve the model. Especially,~ a distance measure formula is proposed to analyze error resulting from different order and the Coxian distribution with optimal order is found.Simulation with Matlab software indicates that the outcome is good.