Sliced inverse regression (SIR) is a clever technique for reducingthe dimension of the predictor in regression problems, thus avoidingthe curse of dimensionality. There exist many contributions on variousaspects of the performance of SIR. Up to now, few attention hasbeen paid to the problem of choosing the number of slices within theSIR procedure appropriately. The aim of this paper is to show thatespecially the estimation of the reduced dimension can be stronglyinfluenced by the chosen number of slices.2000 Mathematics Subject Classification: 62H12
展开▼