directadjusting-package: directadjusting: Directly Adjusted Estimates
Description
Compute estimates and confidence intervals of weighted averages quickly and easily. Weighted averages are computed using data.table for speed. Confidence intervals are approximated using the delta method with either using known formulae or via algorithmic or numerical integration.
Arguments
Recommended installation
devtools::install_github(
"FinnishCancerRegistry/directadjusting@release"
)
Example
# suppose we have poisson rates that we want to adjust for by age group.
# they are stratified by sex.
set.seed(1337)
offsets <- rnorm(8, mean = 1000, sd = 100)
baseline <- 100
sex_hrs <- rep(1:2, each = 4)
age_group_hrs <- rep(c(0.75, 0.90, 1.10, 1.25), times = 2)
counts <- rpois(8, baseline * sex_hrs * age_group_hrs)# raw estimates
my_stats <- data.table::data.table(
sex = rep(1:2, each = 4),
ag = rep(1:4, times = 2),
e = counts / offsets
)
my_stats[, "v" := my_stats[["e"]] / offsets]
print(my_stats)
# sex ag e v
# <int> <int> <num> <num>
# 1: 1 1 0.08928141 8.759527e-05
# 2: 1 2 0.10054601 1.175523e-04
# 3: 1 3 0.11987410 1.238776e-04
# 4: 1 4 0.09722692 8.365551e-05
# 5: 2 1 0.18043221 1.937844e-04
# 6: 2 2 0.14781448 1.227479e-04
# 7: 2 3 0.21747515 1.987203e-04
# 8: 2 4 0.21519746 1.781152e-04
# adjusted by age group
my_adj_stats <- directadjusting::directly_adjusted_estimates(
stats_dt = my_stats,
stat_col_nms = "e",
var_col_nms = "v",
conf_lvls = 0.95,
conf_methods = "log",
stratum_col_nms = "sex",
adjust_col_nms = "ag",
weights = c(200, 300, 400, 100)
)
print(my_adj_stats)
# Key: <sex>
# sex e v e_lo e_hi
# <int> <num> <num> <num> <num>
# 1: 1 0.1056924 3.474049e-05 0.09474912 0.1178996
# 2: 2 0.1889406 5.237509e-05 0.17527556 0.2036710
News
News for version 0.6.1
directadjusting
DESCRIPTION and documentation fixes.
News for version 0.6.0
directadjusting
First CRAN release.
News for version 0.5.0
directadjusting::direct_adjusted_estimates
directadjusting::delta_method_confidence_intervals made a lot more
flexible. It now accepts via conf_method a string, a call, and a list
of calls that produce the desired confidence intervals.
directadjusting::direct_adjusted_estimates
directadjusting::direct_adjusted_estimates option conf_methods = "boot"
removed. Only delta method confidence intervals now possible. Making use of
the delta method is now more flexible and accepts e.g.
list("log", list(g = quote(qnorm(theta)), g_inv = quote(pnorm(g)))).
directadjusting::direct_adjusted_estimates
directadjusting::direct_adjusted_estimates now allows for the sake of
convenience to be called with no adjust_col_nms defined. This results
in no adjusting.
News for version 0.4.0
directadjusting::direct_adjusted_estimates
directadjusting::direct_adjusted_estimates now correctly uses
the same conf_lvls and conf_methods for all statistics when their
length is one.
News for version 0.3.0
directadjusting
Remove deprecated directadjusting::direct_adjusted_estimates. Use
directadjusting::directly_adjusted_estimates.
Author
Maintainer: Joonas Miettinen joonas.miettinen@cancer.fi (ORCID)
See Also
Useful links: