DemogAnr 0.2.0
lifetable() gains a nax_method argument.
In addition to the default Coale-Demeny young-age factors with
n/2 elsewhere ("cd"), it can now refine the
average person-years lived in adult intervals with the iterative Keyfitz
(1966) method ("keyfitz"). A user-supplied nax
vector can also be passed to borrow separation factors from any external
model life table system (e.g. Coale-Demeny or United Nations). Default
behaviour is unchanged.- New
chm_brass(): indirect estimation of child mortality
from children ever born and children surviving classified by age of
mother, using the Trussell variant of the Brass method (United Nations
Manual X, 1983). Supports all four Coale-Demeny model families (North,
South, East, West), returns q(x) for x = 1, 2, 3, 5, 10, 15, 20 with
reference periods and dates. Reproduces the Manual X Panama (1976)
worked example to the published precision.
- New
standardize(): compares two populations by age
standardization following Preston et al. (2001, Chapter 2), using the
average of the two age compositions as the default standard and
reporting the crude and age-standardized rates plus the comparative
mortality ratio (CMR). Reproduces their Box 2.1 (Sweden vs Kazakhstan)
exactly.
- New
decompose_rates(): Kitagawa decomposition of a
difference between two crude rates into rate and age-composition
components.
- New
decompose_LE(): Arriaga decomposition of a
difference in life expectancy at birth into age-specific direct and
indirect/interaction effects.
- Visual output:
dem.fert, dem.cdr,
lifetable, project_population,
decompose_LE, myers, whipple and
un_age_sex_accuracy gain a graph = TRUE
argument. When enabled, a ‘ggplot2’ plot of the analysis is attached to
the result ($plot) and shown when the object is printed;
plot() methods are provided for each. Scripts that assign
the result draw nothing, so the behaviour is safe in batch code.
- New
pyramid(): draws a population pyramid (back-to-back
bar chart by age group and sex) with ‘ggplot2’.
- New age-quality / age-heaping indices:
whipple()
(Whipple’s index of digit preference for 0 and 5), myers()
(Myers’ blended index over all ten terminal digits),
sex_ratio() and age_ratio() helpers, and
un_age_sex_accuracy() (the United Nations joint age-sex
accuracy index). un_age_sex_accuracy() reproduces the Ghana
1960 worked example of Kpedekpo (1982); Myers follows Rodriguez (2015) /
Myers (1940).
- New bundled data set
ghpop2021: Ghana 2021 single-year
age distribution by sex, for the digit-preference indices.
- New bundled data set
ghmort2021: age-specific
population and deaths for Ghana in 2021 by sex, for demonstrating
sex-specific life tables and the decomposition of the male-female gap in
life expectancy.
DemogAnr 0.1.0
- Initial CRAN release.
- Added functions for basic demographic analysis: fertility
(
dem.fert), mortality (dem.cdr,
dem.mmr, dm.chm), and life tables
(lifetable, lifetable_nqx).
- Implemented mathematical population projection
(
math_project), a stochastic simulation-based projection
(project_population), and demographic interpolation models
(interpolation, karup_king,
brass_logit).
- Addressed CRAN reviewer feedback:
- Formatted description and references in the
DESCRIPTION
file to match CRAN rules.
- Suppressed all raw console printing in core demographic functions
and implemented custom S3 print methods (
print.dem_cdr,
print.dem_fert, print.dem_mmr, and
print.dem_chm) to align with standard R package
behavior.
- Added academic references to the documentation blocks of all core R
files.
- All invalid links have been removed from the documentation and
Description file.
- Methodological corrections prior to release:
brass_logit() now uses the correct inverse of the Brass
half-logit transform, exp(2Y)/(1 + exp(2Y)), and documents that its
input is the cumulative probability of dying q(x) = 1 - l(x).lifetable_nqx() and lifetable() now
account for the width of each age interval (1, 4, 5, … years) when
computing nqx and person-years lived (nLx); lifetable()
gains sex and radix arguments and uses
Coale-Demeny separation factors for ages 0 and 1-4.karup_king() defaults now work out of the box (packaged
Karup-King coefficient matrices).project_population() now propagates all simulated
trajectories through the projection horizon and reports lower quartile,
median, mean, and upper quartile per year; net out-migration now reduces
growth.dm.chm() neonatal filter now uses age < 1
month.- Removed unused dependencies (tidyverse, brms, logger, data.table)
and leftover Stan scaffolding.