|
Abstract:
|
Multiple heterogeneous integrative study is very challenging in high-dimensional integrative settings, especially when data have heavy-tailed distribution or outliers exist in random errors and covariates. Under ultra-high dimensional sparse regression models, we propose a novel robust integrative estimation procedure by aggregating local high-dimensional redescending M estimators in this paper. In theory, we provide some sufficient conditions under which the aggregated redescending M estimators possesses consistent variable selection result. The finite-sample performance of the proposed procedure is studied via extensive simulations and two real data integrative studies.
|
