通过利用Shi,Shi-Jiang算法和一个Bootstrap抽样技巧,给出了一个推广了的序约束推断方法,在不需要满足方差恒定假设的前提下处理相同的选择和聚类问题.模拟结果显示此方法比Peddada和Simmons-Peddada方法更好地控制了假阳性率.
Via the application of Shi and Shi-Jiang algorithms and a Bootstrap sampling technique,it proposed a generalized order-restricted inference methodology for the same selecting and clustering problem without satisfying the homoscedasticity assumption.Simulation results show that the procedure can control the false positive rate better than that developed by Peddada et al and Simmons and Peddada.