Random Weighting Method for Estimating Sampling Distributions

This paper presented a new random weighting-based sequential variance minimization method for estimating the sampling distribution of a statistic. The optimal resampling probability weight was calculated via direct variance minimization and the random weighting estimation of sampling distribution was constructed by using the obtained optimal importance resampling distribution through the cross-entropy procedure. The convergence of the proposed method was also rigorously proved. Simulations analysis demonstrates that the proposed method can effectively estimate the sampling distribution of a statistic, resulting in much higher estimation accuracy than the existing methods.