COINr 1.1.14
- Minor bug fixes: spaces in uCodes, issue exporting 2-level composite indicators to Excel
- Add scaling parameter to
n_scaled()
- Option to disable outlier treatment for purses
COINr 1.1.12
- Enable custom operations on coins and purses: new generic
Custom()
with methods for coins and purses.
- Enable different
by_df
and w
by aggregation level in Aggregate.coin()
- Add possibility to impute panel data using “constant” method from
stats::aprox()
.
COINr 1.1.11
- Allow retention of units with no data in split to purse
- Allow NAs in group vector for
i_median_grp()
- Enable linear interpolation in panel data with
impute_panel()
- Updated documentation on aggregation and imputation functions
COINr 1.1.9
- Fixed results tables which rounded before converting to ranks, resulting in precision errors
COINr 1.1.8
- Added possibility to set normalisation parameters in
iMeta
, added more explanation and examples in the normalisation vignette.
- Fixed bug correlation directions
COINr 1.1.7
I have skipped a number of version number iterations. Here’s what’s changed since last time:
- Add some helper functions to convert codes to names
- Add flip axes to bar chart
- Add option to skip SA address checks
- Option to use raw data in results tables
- Small bug fixes
- Catch more errors on coin construction
COINr 1.1.2.9000
Dev version of COINr until next CRAN submission!
Treat()
now catches errors - data frame/coin/purse methods
COINr 1.1.2
Lots of small fixes and feature extensions, including
- More robust write to log, can deal with
COINr::func_Name
type calls
- Catch spaces and numbers in iCodes
- Tweaks to correlation tables
- Reordering for correlation plots
- Better input checks
- Extra stats
- Bi-direction log transformation for negative skewed outlier treatment
- Subgroups of units for bar charts
COINr 1.1
Some new functions and bug fixes:
- Removed defunct function messages and startup message, defunct functions are now removed from the namespace
- New
get_trends()
function for analysing time trends
- Option to export purse class to Excel
- Expand unit test coverage to 80%
- Quite a lot of small bug fixes
- Improved documentation in many places
- Improved colouring for some plots
- Enabled distance to target normalisation plus examples in vignette
- COINr is now citable by a JOSS paper
COINr 1.0.0
Major update to COINr. The main changes are as follows:
- Syntax changed to be more consistent
- New streamlined “coin” class replaces older “COIN” class
- New “purse” class to deal with panel data (time dependent data)
- Removed interactive functions based on plotly and shiny
- Much fewer package dependencies
- Underlying code rewritten to be more robust
- Improved functionality and flexibility in many functions
- COINr 0.6.1.9000 has been archived as a separate package called “COINr6” - this can be installed to ensure older code will still run.
The full changes are too extensive to write here. See vignette("v1")
also available online here for a more complete description of changes.
Note that these changes are disruptive but greatly improve the package and make it more maintainable for the future. This is a one-off major overhaul, after which backwards compatibility will be ensured.
COINr 0.6.2
- Fixed bug in
aggregate()
which was wrongly assigning weights in some circumstances (note this is a fairly significant bug - worth re-running your results as it may have affected them)
- Added colour option to
plotframework()
- Fixed bug in
compareDF()
(resulting from tibbles vs data frames)
- Fixed bug in
sensitivity()
causing trouble with running SA with no weights uncertainty specified
COINr 0.6.1
- Removed dependency on new native pipe so now runs again on R > 4.0
COINr 0.6.0
- New plot function
plotIndDot()
for plotting single indicators with possibility to label units
- Added
print
method for COIN class. More methods to come.
- Extended support for panel data in
assemble()
. This can now output multiple COINs structured into a tibble. The idea of working with multiple COINs will be extended in future versions.
- Improved bug trapping in Winsorisation function
- Added data availability thresholds during aggregation - see
aggregate()
function.
- Better NA colour for
plotCorr()
plus fixed variable ordering
- Further colour options in
plotCorr()
- Variable/aggregate ordering in
plotCorr()
now matches grouping in higher levels
- Add option to do PCA not in groups in
getPCA()
- Colour options for
plotSARanks()
- Added dominance pairs calculation to
ourankMatrix()
- Stopped max tab length exceeding in Excel export.
- Online book now pretty much up to date with COINr 0.6.
COINr 0.5.5
- COINr now requires R version 4.0.0 or above - this is following some issues on some distributions during CRAN checks.
- Improved documentation and bug catching in
aggregate()
- New function
replaceDF()
which is a simple utility for multiple values in one go for a data frame.
- New function
removeElements()
which successively removes indicators or aggregates one at a time and summarises the effects.
COINr 0.5.4
- Fixed various small issues that were flagged by CRAN
- New function
extractYear
which allows panel data to be passed to assemble()
and (optionally) imputed by latest year.
- Fixed bug in
plorCorr()
and getCorr()
- Changed
WorldDenoms
data set slightly to only use ASCII coding - this involved unfortunately removing a few accents on country names which is necessary to avoid warning notes when submitting to CRAN.
- Added goalpost normalisation method, as in European Skills Index
- Fixed small bug in
compareDF()
COINr 0.5.3
Fixes following comments from CRAN.
- Patched up holes in the documentation
- Function examples now run where possible
- Shiny apps only run now in interactive mode
COINr 0.5.2
This is the first release submitted to CRAN. Changes include:
- Documentation has been thoroughly cleaned up, including coherent examples for each function.
- Small bugs have been checked and fixed.
- A vignette is now available