biasAdjustment: Bias adjustment

In schwenzfeier/udp: Statistically Useful Differential Privacy

Description

Calculates \tilde θ^{dp} from \hat θ^{dp}, which is biased due to censoring.

Usage

biasAdjustment(theta_dp, alpha, lambda, upper, two_sided)

Arguments

theta_dp	Mean of censored QOI returned from BLB with dp noise added
alpha	Primary alpha. Differentially private estimate of censoring on one side of the distribution of theta.
lambda	Bounding parameter
upper	Indicator for whether the alpha represents upper (right-sided) or lower (left-sided) censoring
two_sided	Indicator for whether censoring is taking place on both sides or just one of the distribution of theta

Value

theta_tilde	Bias corrected QOI
sigma	Estimated sigma
alpha	Estimated secondary alpha
two_sided_ba_ind	Indicator for whether one or two-sided bias adjustment was used (just for debugging, not differentially private)

schwenzfeier/udp documentation built on April 1, 2021, 3:39 a.m.