LearnNonparam 1.3.0
- R side
- Added several two-sample distribution tests:
distribution.ks, distribution.kuiper,
distribution.cvm, distribution.ad
- (Breaking) Removed
twosample.scoresum
and twosample.ks
- (Breaking) Modified two parameters in
define_pmt: renamed inherit to
method; updated the rejection parameter to
accept only "<>", "<", or
">"
- Added
quickr support in define_pmt
- Introduced an
autoplot method for the
PermuTest class
- Fixed an issue where p-value calculations were affected by
double-precision floating-point errors
- Resolved an edge-case issue in
RCBDTest that caused
errors during score computation
- C++ side
- Implemented a new progress bar with additional features: speed and
ETA
- Significantly improved the efficiency of calling R functions from
C++
- Enhanced permutation efficiency in two-sample and paired two-sample
tests using branchless swap techniques
- Removed bounds checking during permutation
- Fixed an issue in
define_pmt where
statistic could not be a mutable lambda
LearnNonparam 1.2.9
- Fixed incorrect permutation statistic computation due to deep copy
behavior of
Rcpp::Language
LearnNonparam 1.2.8
- R side
- Refined the calculation of confidence bands in
CDF
- Added DKW inequality-based confidence bands in
CDF
- Enhanced error handling in
define_pmt
- C++ side
- Rolled back
twosample.* functions
LearnNonparam 1.2.7
- R side
- Fixed p-value calculation and adopted the correlation coefficient as
test statistic in
Correlation
- Added display of test types when printing user-defined tests
- Refined the compilation of R functions in
define_pmt
- Improved calculating efficiency in
PairedDifference
- C++ side
- Enhanced the permutation efficiency in
paired.*
- Reduced the number of permutations in
twosample.* in
cases with duplicate elements across samples
LearnNonparam 1.2.6
- R side
- Fixed the issue with parameter passing to
ggplot2::stat_bin in plot()
- Added detection for missing values and zero-length data in
test()
- Switched to version 3 serialization format
- Added more documentation
- C++ side
- Improved the efficiency of calling R from C++
- Enhanced the permuting efficiency in
rcbd.*,
association.*, and table.*
- Replaced
R_len_t with R_xlen_t to support
long vectors
- Replaced
Rcpp.h with Rcpp/Lightest to
reduce compilation time
LearnNonparam 1.2.5
- Added support for
scoring as an active binding in
define_pmt
- Corrected x-axis range in
plot() when
style = "graphics"
- Improved the performance of
PairedDifference
LearnNonparam 1.2.4
- R side
- Added support for custom scoring systems in
define_pmt
- Enabled compatibility with C++ standard versions beyond C++14 in
define_pmt
- Improved the performance of
KruskalWallis,
OneWay, and Studentized
- Replaced certain
warning() calls with
stop()
- Refined some text
- C++ side
- Enhanced permuting efficiency in
twosample_pmt
LearnNonparam 1.2.3
- Improved
twosample_pmt
- Added examples
- Fixed
SiegelTukey and
ContingencyTableTest
LearnNonparam 1.2.2
- R side
- Changed the default value of
n_permu to
1e4
- Added support for test statistics defined using Rcpp in
define_pmt
- Added support for
null_value other than zero in
two-sample location tests
- Added support for
scoring in
TwoSampleAssociationTest
- Added a confidence interval for p-value in
print()
- Added the
LearnNonparam.pmt_progress option
- Added
attr(t$n_permu, "n_used")
- Replaced
multicomp.* with multcomp.*
- Replaced
KSampleF and RCBDF with
OneWay and RCBDOneWay
- Stopped exporting abstract classes (
TwoSampleTest,
KSampleTest, etc.)
- Improved the efficiency of
ChiSquare,
Difference, JonckheereTerpstra,
KruskalWallis, RatioMeanDeviance and
Studentized
- Fixed many bugs
- C++ side
- Introduced a new progress bar, which is built at compile time to
minimize runtime overhead
LearnNonparam 1.2.1
- R side
- Added
define_pmt
- Added some error messages
- Added some active bindings
- Introduced
graphics as an alternative to
ggplot2
- Introduced
compiler::cmpfun for better performance
- Renamed
ANOVA, RCBD,
RCBDANOVA and SignedDiff
- Replaced
MultiCompT and TukeyHSD with
Studentized
- Replaced
"approx" with "asymp"
- Improved
MultipleComparison
- Improved
CDF$plot()
- Fixed many bugs
- C++ side
- A new progress bar
- Support for larger
n_permu
LearnNonparam 1.2.0
- Added logo
- Introduced Rcpp
- Replaced
Mean with Difference
- Replaced
SignedScore with SignedDiff
- Replaced
PermuTest$feed() with
PermuTest$test()
- Accelerated
RCBD and
TwoSamplePairedTest
- Improved
print() and plot() method of
CDF, MultipleComparison and
PermuTest
- Added
"approx" option for the
Quantile$type
- Fixed many bugs