CFE 2019: Start Registration
View Submission - CMStatistics
B1624
Title: Sample-specific stability selection with effective error control Authors:  Heewon Park - Hiroshima University (Japan) [presenting]
Abstract: Recently, sample-specific analysis has drawn a large amount of attention for identifying patient-specific characteristics in the progression of cancer. In order to effectively identify sample-specific molecular mechanisms, we propose a novel sample-specific feature selection method based on the stability selection. Although stability selection provides effective results for variable selection, the method's results are sensitive to the value of the regularization parameter because the method performs feature selection based only on the particular parameter value that maximizes the selection probability. To settle on the issue, we propose robust stability selection and show that our method provides an effective theoretical property (i.e., per-family error rate). We then develop a sample-specific stability selection method based on the kernel-based L1-type regularization and weighted random re-sampling technique. The proposed method estimates the selection probabilities of variables using the sample-specific random lasso and then perform feature selection based on robust stability selection. We observe through the numerical studies that our strategies can effectively perform sample-specific analysis.