摘要
目的探讨通过Bootstrap抽样的方法来解决决定系数的可信区间估计。方法通过实例分析,重复从原始数据中进行有放回的抽样得到Bootstrap样本,然后计算每个Bootstrap样本的决定系数的Bootstrap估计量。结果Bootstrap估计量的BCa可信区间不仅矫正了非对称性,还对原始数据的异常值给予了矫正。结论 Bootstrap抽样不需要任何理论推导,使用方便,其估计量的BCa可信区间能够自动校正统计量的非对称性,因此更具优良性。
Objective To explore the Bootstrap sampling method to solve the confidence interval estimation of the coef- ficient of determination. Methods Through case analysis, obtained Bootstrap samples through repeatedly returned sampling from the original data, then calculated the Bootstrap estimation of the coefficient of determination of each Bootstrap sample. Results The Bootstrap estimation＇s BCa confidence interval not corrected asymmetry,but also corrected the outer values. Conclusion The Bootstrap sampling did not need any theoretical derivation, use convenient, and the BCa confidence interval of the Bootstrap estimation was more advantages.
出处
《中国卫生统计》
CSCD
北大核心
2014年第1期49-52,共4页
Chinese Journal of Health Statistics
基金
国家自然科学基金(No:81172774)
国家自然科学基金青年基金(No:81001294)
江苏省高校哲学社会科学基金项目(No:2013SJB790059)
