A***A 发帖数: 98 | 1
Benefit:
Second order accuracy (see Peter Hall's papers).
Less restrictive distributional assumptions, as is the case with all
resampling based methods.
Simplicity! Yes, simplicity for complicated problems.
Uses: a lot of uses, here is a tentative list
Esimation:
est. bias
est. locationa nd dispersion
Confidence sets
Hypothesis testing
Regression and all kinds of linear models or nonlinear models
Forecasting and time series
Well. The Bootstrap does not always yield narrower CI than a norm |
|