NEWS

anthro 1.1.0.9000

The Weight-for-length/height (zwfl) z-score indicator is now also computed for observations where weight or lenhei is outside the plausible ranges. The plausible ranges were 0.9-58 for weight and 38-150 for lenhei.

The resulting data frame of anthro_zscores has a new column c9mo_flag to indicate whether an incorrect measurement position was observed with children younger than 9 months.

For inputs with cluster/strata = NULL a faster method was added for this special case to compute anthro_prevalence. This method yields the same results within an observed tolerance of < 0.0001 on test data.

Fixed a bug during prevalence computation when a value used as a group (like sex) contains NA and at the same time all values in that group are NA as well. Previously this resulted in a hard stop of the computation, now the code gracefully handles it.

Improved prevalence computation code, it now consumes slightly less time and is better organized. In addition test-coverage and documentation is improved.
Cluster/strata/sw values in anthro_prevalence now support an explicit NULL value to better control for (missing) cluster/strata/sw information.
New API functions are exported, mainly useful to use complicated portions of the logic in other packages, not meant for end-users.
nest = TRUE is now set for the survey designs in anthro_prevalence. See the documentation of survey::svydesign for more information.
Removed headc, armc, triskin and subskin from parameters of anthro_prevalence as they were not used for the final output.
_pop/unwpop values in prevalence calculation are now 0 instead of NA if no values are present for that group.

Previously, if one level in a group only has only one observation, all stdev columns in anthro_prevalence are NA. Now only the levels with one observations yield NA, for the others the stdev is still computed.

Handling of age information is improved and loss of information from converting from age in days to age in months or vice verse is reduced.
Some code improvements.

Fixed a bug where, observations where erroneously removed from prevalence computation. This happens if age in months was supplied, then values > 59.992 months were considered > 1826 days. Now anything above 60.009 months is considered > 1826 days and thus excluded.

Z-scores are now only computed for age < 60 months instead of age <= 60 months.
Z-scores for wfl were previously also computed when age >= 60 months and <= 1856 days.

Age in days is now rounded half to even (e.g. 730.5 days = 731) before joining the data with the reference values. This is in line with previous implementations and in particularly relevant for data points with age exactly 24 months. Previously the example above was converted to 730 days and with this release it is converted to 731 days (#17).
The cleaned measure and sex variables are now part of the output data.frame of anthro_zscores (#20, #24).
Removed covr from suggests dependencies.
Fixed a typo in the docs (#15).