LearnNonparam v1.3.0

• 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 v1.2.9

• Fixed incorrect permutation statistic computation due to deep copy behavior of Rcpp::Language

LearnNonparam v1.2.8

• 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 v1.2.7

• 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 v1.2.6

• 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 v1.2.5

• 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 v1.2.4

• 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 v1.2.3

• Improved twosample_pmt
• Added examples
• Fixed SiegelTukey and ContingencyTableTest

LearnNonparam v1.2.2

• 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 v1.2.1

• 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