| Summary: | 碩士 === 淡江大學 === 數學學系碩士班 === 101 === The regression diagnostics that assess the adequacy of a regression model for observed data should be conducted before the regression analysis. For complete or right censored failure time data, several graphical or quantitative methods have been established for regression diagnostics (e.g. Collett (1994), Klein and Moeschberger (2003), and Lawless (2003)). In the thesis, we develop a diagnostic method for assessing the proportional hazards and proportional odds assumptions with current status data under the odds-rate model (Scharfstein, Tsiatis, and Gilbelt (1998)) and provide a stable and efficient computation method for proposed estimators. We illustrate the performance of our method via simulation studies and three real data analysis.
|
|---|
