Bootstrap和Logistic Regression在医学统计中的应用

The Application of Bootstrap and Logistic Regression in Biostatistics

将Bootstrap和Logistic Regression相结合,解决医学统计中因样本量小造成的统计结果的可信性问题.首先对一组医学统计数据应用Logistic Regression进行参数估计和假设检验,然后应用Bootstrap重新对模型和参数进行假设检验.算例表明,根据Bootstrap法得到的统计结果能对由Logistic Regression回归模型得到的结果的可信性做出判断.

The Bootstrap and logistic regression are combined to solve the credibility problem arising from the small samples in biostatistics. First, the Logistic regression model is applied to a categorical data set in biostatistics to get the parameter estimations and hypothesis tests. Then the Bootstrap is applied to get the new hypothesis tests in both the model and the parameter estimations. The test example shows that the outcomes from the Bootstrap method can help us to make a decision for the ones from the logistic regression model.