摘要
应用非参数Bootstrap方法对Monte-Carlo仿真输出样本进行重复再抽样,用一种改进的偏差校正方法给出了命中概率的点估计,用BCa方法对命中概率进行置信区间估计,并使用传统统计分析方法对Monte-Carlo仿真输出样本进行分析,将两种方法得出的分析结果进行对比。比较结果表明,Bootstrap估计方法在置信区间估计方面有较大优势,能够降低对Monte-Carlo仿真输出样本容量的要求。
The output from Monte-Carlo simulation was replicated using Non-Parameter Bootstrap method. The estimate of hitting probability using an improved bias-corrected estimator and the confidence interval based on BCa estimator were compared with those from traditional estimators on analyzing Monte-Carlo simulation output. It is shown that Bootstrap estimators provides shorter confidence interval and requires fewer samples from Monte-Carlo simulation output.
出处
《系统仿真学报》
CAS
CSCD
北大核心
2010年第6期1347-1349,共3页
Journal of System Simulation
