A0685
Title: Analysis of gene expression data subject to measurement error in binary responses and predictors
Authors: Li-Pang Chen - National Chengchi University (Taiwan) [presenting]
Abstract: Gene expression variables are usually used to classify specific diseases, such as acute myeloid leukemia (AML) and acute lymphoblastic leukemia (ALL). However, gene expression data usually encounter ultrahigh dimensionality and measurement error. Those complex features also affect the classification performance. The aim is to introduce a method called BOOME, which refers to BOOsting algorithm for measurement error in binary responses and ultrahigh-dimensional predictors. This method primarily focuses on logistic regression and probit models with responses and predictors contaminated with measurement error. The BOOME method aims to address the effects of measurement error and then employs a boosting procedure to make variable selections and estimations. Numerical experiments reveal that the BOOME method is valid for addressing measurement error effects and deriving reliable estimation results.
