Changelog • chartplotter

Sets colors to green for new D-score references (despite their WHOpink/blue defaults)

Solves a bug in the selection of the preterm reference in set_curves()

Updates select(), drop_na() and rename() function to tidyselect 1.2.0 grammar that prohibits names like .data$x
Adds extra sorting operation on age to so that WFH points are always connected in their time sequence

Repairs bug introduced by new behavior of predict() from brokenstick 2.3.0

Replaces “tibble with attribute” by list with elements psn and xyz

Replaces new_data by newdata for brokenstick 2.0.0 package
Removes format = 1 arguments for read_bds() in tests
Update dependency versions to brokenstick 2.0.0 and donorloader 0.31.1

Use data from the jamesdemodata rather than bdsreader package

Supports the new format arguments in bdsreader::read_bds()

Solves problem with chart without ynames
Adds some tweaks to GitHub Actions scripts

Relocates dependencies to growthcharts organisation

Relocates back to stefvanbuuren account
Adds GHA R-CMD-check and pkgdown

Relocates dependencies to growthcharts organisation
Improves handling of empty target object

MAJOR UPDATE
Replaces jamestest by jamesdemodata package
Replaces minihealth by bdsreader package
Update all code and test files

Updates ind.rds to minihealth 0.90.0
Adds a test on testfile 5

Main release, now completely based on centile package
Unexport functions for internal use
Switch to markdown documentation
Solves problem with curve interpolation of WFH
Limit prediction horizon to 2 years when there is no model
Sets the maximum prediction horizon equal to last internal knot
Solves a problem with curve interpolation of the D-score
Evades prediction on NULL models
Silences pivot_longer() if no ynames are found
Makes yname optional so that it doesn’t get stuck on dsc
process_chart() returns early if there are no data
Styles the R sources

Simplifies plotting logic
Solves the problem that did not plot plot T 3254 and T 6021
Solves bug that did not plot 1-point curve
Adds data dump for debugging purposes
Add safe_approx() function to deal with zero observations
Uses rule = 2 in approx to calculate Z-score outside the reference (note: this may not always be appropriate)

Repair typo in set_curves() that crashes JAMES
Make set_curves() work with zero observations

Replace yzy by centile package

Major overhaul
Splits calculation and plotting
Replaces clopus dependency by yzy and jamesxzy imports

Accepts new dnr arguments: "0-2", "2-4" and "4-18" as donor data choices
Repairs the curve interpolation problem for Kevin S

This version major extension of the chartplotter package.

Shifts all calculations to the Z-score analysis metric by transform_z()
Explicit definition of the display metric depending on target individual’s sex and ga by transform_y()
Renames “.z” to “_z” variables
Improves various plotting functions

Major update: Consistent use of measurement-analysis-display metrics
Prediction lines not yet functional

Repairs discrepancy in Z-transformation in find_matches() that resulted in incorrect Z-scores for preterms in mixed data (#2)

Removes grid.draw() from process_chart(), so now your need to run grid.draw() to render the graph

Reinstates dependency brokenstick::get_knots() because minihealth needs the brokenstick dependency anyway (for predict)

Removes dependency on brokenstick::get_knots() by making a copy

Updates to R 4.0.0 and more recent brokenstick, curvematching packages

Draws prediction line as continuous between last observation and prediction
Adds a NEWS.md file to track changes to the package.