A0952
Title: Fast computation of the bootstrap method for incomplete data
Authors: Masahiro Kuroda - Okayama University of Science (Japan) [presenting]
Abstract: The bootstrap method is a powerful tool for making statistical inference about a parameter of a statistical model. The bootstrap generates a sample by randomly sampling with replacement from the observed data. The maximum likelihood estimate (MLE) of the parameter is computed from this sample. By repeating the bootstrap sampling procedure, the parameter distribution can be obtained. When dealing with incomplete observed data, an iterative computation step is required to find the MLE in the bootstrap procedure. The EM algorithm is used in this step and applied to each of the bootstrap samples. However, the bootstrap can be time-consuming due to the slow convergence of the EM algorithm. To address this issue, a fast bootstrap method is proposed that includes an acceleration step to speed up the convergence of the EM algorithm. Numerical experiments apply the proposed bootstrap method to contingency table analysis and examine its performance in terms of the number of iteration and CPU time.
