P-Ⅲ型分布参数估计方法的比较研究

Comparative Study on Parameter Estimation Methods of Pearson Type Ⅲ Distribution

应用统计试验对P- 型分布的几种参数估计方法的无偏性和稳健性做了分析比较研究,包括矩法、概率权重矩法、数值积分单权函数法、数值积分双权函数法、混合权函数法和线性矩法。结果表明数值积分单、双权函数法的无偏性较好,线性矩法次之;线性矩法的稳健性最好,混合权函数法次之,矩法最差。

The statistical experiment was applied to compare six different parameter estimation methods for Pearson type Ⅲ distribution, including conventional moment (MOM),probability weighted moment (PWM),numerical integral with singleweighted function (NISWF),numerical integral with doubleweighted function (NIDWF),mixed weighted function moment (MWFM) and Lmoment (LM). The statistic test based on unbiasness and robustness properties was designed and performed. Experimental results show that NIDWF and NISWF are the best for unbiasness,while LM is better than the others;for the robustness,LM is the best and MOM is the worst method, while MWFM is superior to other methods.