Quantile Estimation Using the Maximum Transformation, Stochastic Approximation and the Jackknife.
NAVAL POSTGRADUATE SCHOOL MONTEREY CALIF
Pagination or Media Count:
The rate of convergence of the expected value of quantile estimates using a stochastic approximation with the maximum transformation is evaluated. The analysis is performed using linear regression techniques on computer simulation results for quantile estimates of the unit exponential distribution. Included is a discussion of the use of the jackknife technique to reduce the bias of the stochastic approximation quantile estimates. Simulation results for the 2-fold jackknife for the m sup-.5 term are tabulated. The main conclusion of the analysis is that the lowest order term in the expression for the expected value of the estimate as a function of sample size decreases as m sup-.25. Author
- Statistics and Probability