RisksetsToIpdSkewed: Expand weighted risk set data with variance to long format...

In kaz-yos/distributed: Examine Privacy Preserving Data Analysis Methods in Simulated Distributed Data Network

Description

Expand weighted risk set-level data to individual-level data or summary long-format data. The weights are regenerated to maintain the given mean and variance. This version specifically try to avoid negative weights by skewing distribution. Also computation is based on the summary version, which is expaned when compress = FALSE. The individul-level data contain one row for each individual in each risk set. The summary long-format data are compressed so that each row represent multiple individuals. The count variable indicates how many individuals each row represent.

Usage

RisksetsToIpdSkewed(x, compress, helper_fun = RisksetsToIpdSkewedHelper)

Arguments

x	data frame generated by RequestSiteRisksets, containing risk set-level data.
compress	defaults to FALSE. If true, the summary long-format data containing one row for each unique combination of variables including weights are created. The count variable indicates how many individuals each row represent.
helper_fun	helper function used to regenerate weights given sample size n, sample mean m, sample variance v, and labeling for treatment A and event status event. Defaults to RisksetsToIpdSkewedHelper, which tries to balance the number of observations with above-mean and below-mean weights. RisksetsToIpdExtremeHelper will positions just one observation with an above-mean weight and all others below mean.

Value

data frame populated with point estimates and variance estimates by various methods.

Author(s)

Kazuki Yoshida

kaz-yos/distributed documentation built on May 27, 2019, 4:50 a.m.