On quantile estimation by bootstrap
Artikel i vetenskaplig tidskrift, 2006
Exact bootstrap is used to optimize the weights of an L-estimator for quantiles with respect to the estimated MSE (mean square error). Performance of the new estimator is measured by comparing MSE with the sample quantile. The new estimator performs better than the sample quantiles in almost every case. However, the gain is only about 5%, in terms of decreased MSE.