|
SEMINARS |
Principle Seminar of the Department of Probability Theory, Moscow State University
|
|||
|
An improvement method of sliced inverse regression using principal component analysis T. Akita Hiroshima University |
|||
Abstract: We suggest an improvement method of sliced inverse regression (SIR) which is called gPCA-SIR h. The idea of the method is re-separating the group of the data, which is separated in some slices of the response variable, by using principal components. We also compare SIR with PCA-SIR by simulation, and prove the convergence of estimator. |