Nonparametric confidence bands construction for GLM models with length biased data
Jorge L. Ojeda, José Antonio Cristóbal Cristóbal, José Tomás Alcalá Nalvaiz · Journal of nonparametric statistics · 2004
In this work, we propose a nonparametric confidence bands construction for the linear predictor in generalized linear models (GLM) when data is affected by length bias. These confidence bands are suitable to keep track of the behavior of the linear predictor for a given GLM family and link. To that end we adapt the result of Bickel, P. J. and Rosenblatt, M. (1973) about the supremum for stationary Gaussian processes.