Confidence Curves in Nonlinear Regression
针对正态非线性回归模型中Wald置信区间不准确的问题，提出一种基于剖面似然的图形化方法及稳定算法，用于标量参数的推断。
Abstract Standard Wald confidence regions for parameters in a normal nonlinear regression model often fail to capture accurately the uncertainty of estimation as reflected by the corresponding profile log-likelihood. We present a graphical method, along with a stable computational algorithm, for inference on scalar parameters in a nonlinear regression model.