摘要
提出有关列联表面貌的预测和似然比检验的预测误差推理方法,用于单向有序列联表和小样本的预测.其目的是提供预测准确度和高效的分析工具,以及更普遍的预测检验能力.仿真实验显示,该方法可以用于通常的小样本.若似然比检验所得到的P值结论与单向有序列联表的面貌不一致时,应依据基于概率积分变换的0.074校准参数修正P值误差,使P值结论与单向有序列联表的面貌一致.我们将似然比检验的预测误差推理方法应用到胃病的分析,得到证型的胃痛严重度参数与列联表面貌的预测一致的结果.
We develop procedures for inference about the predictions of contingency table profile and the prediction errors of likelihood ratio test,when there is a contingency table where one factor as ordinal variable and with small sample sizes.The aim is to provide tools for analysis of predictive accuracy and efficiency,and,more generally,of predictive ability.Simulations indicate that the procedures can work well in samples of size typically available.If P-value conclusions of likelihood ratio test do not consistent with profile predictions of contingency table,we should revise P-value error of 0.074 calibration parameters based on probability integral translation,make P-value conclusion consistent with profile predictions of contingency table with one factor as ordinal variable.
出处
《厦门大学学报（自然科学版）》
CAS
CSCD
北大核心
2010年第5期612-616,共5页
Journal of Xiamen University：Natural Science
基金
国家自然科学基金(60803078)
关键词
概率积分变换
似然比检验
校准参数
Rosenblatt变换
中心极限定理
probability integral translation
likelihood ratio test
calibrated parameters
rosenblatt transformation
central limit theorem
