Robust versus Non Parametric Approaches and Survival Data Analysis

Catherine Huber-Carol (1)

1 MAP5 - UMR 8145 - Mathématiques Appliquées Paris 5

Abstract There are famous examples of acute sensitivity of optimal parametric procedures to a slight discrepancy from the assumed model. The non parametric approach may also suffer from this kind of deficiency, so that robust procedures (for test and estimation) which are optimal in a neighborhood of a parametric model are to be preferred. This general concept is applied here to survival data analysis.

Keywords: Robustness; Sensitivity; Neighborhoods; Breakdown point; Survival; Influence curve

Domains: Mathematics [math]; Statistics [math.ST]

Catherine Huber-Carol. Robust versus Non Parametric Approaches and Survival Data Analysis. CSAA 2014 International Conference on Statistics and Advanced Applications, Nov 2014, Constantine, Algeria. ⟨hal-01453057⟩