Semi-functional partial linear model with missing at random responses
This work considers a semi-functional partial linear model in the presence of responses that are missing at random. Three estimation approaches are examined: imputation, semiparametric regression surrogate, and inverse marginal probability weighting. For each method, asymptotic properties of the estimators of the model components are derived. A simulation study and an application to real data are used to assess and compare the finite-sample performance of the proposed estimators.
Keywords: missing-at-random responses functional data semi-functional regression partially linear models