Packet 9: Tossup 18

This technique’s asymptotic validity was studied by Peter Hall using Edgeworth expansions. This technique’s “wild” form handles heteroskedasticity, while its “block” form handles autocorrelation, which is used to analyze time series. Like a simpler technique developed (-5[1])by Tukey and Quenouille (“kuh-NOO-ee”), this technique developed by Bradley Efron relies on the plug-in (10[1])principle. (10[1])This technique is done several times to produce a set of decision trees, which are then aggregated, (10[1]-5[1])in a portmanteau meta-algorithm (10[1])for training random forests called bagging. (10[2]-5[1])In this technique that generalizes the jackknife, (-5[1])one samples with (-5[1])replacement (10[2])several times (10[1])from a dataset (10[1]-5[1])and recomputes a quantity (10[1])each time to estimate its original sampling uncertainty. For 10 points, (10[1])name this statistical technique whose name suggests that the data “lifts itself up.” (10[2])■END■ (10[7]0[4])

Back to tossups

Buzzes

Summary