MultivariateStats.jl Versions Save

MultivariateStats v0.10.2

Diff since v0.10.1

Merged pull requests:

• Fix missed AbstractQ tests (#211) (@dkarrasch)
• fix typos (#212) (@spaette)
• CompatHelper: bump compat for StatsBase to 0.34, (keep existing compat) (#218) (@github-actions[bot])

MultivariateStats v0.10.1

Diff since v0.10.0

Closed issues:

• fitting `SubspaceLDA models fails. (#200)
• predict for MulticlassLDA is throwing error (#204)

Merged pull requests:

• Detection of degenerate between-class covariance matrix (#201) (@OkonSamuel)
• MulticlassLDA documentation (#202) (@pnavaro)
• Bug fix & documentation: KernelPCA w/ precomputed X (#208) (@eahenle)
• Matrixify qr.Q instead collecting (#210) (@dkarrasch)

MultivariateStats v0.10.0

Diff since v0.9.1

Closed issues:

• PCA example in the docs doesn't work because DataArray is no longer viable (#120)
• Refresh the documentation (#137)
• Update docs for the new DataFrames version (#151)
• Stable documenter docs are not building (#161)
• PCA example throws error (#166)
• Stable docs link 404-ing (#181)
• Multiclass LDA bounds error (#187)
• Any plan for functional PCA? (#188)
• transform and reconstruct (#190)
• Request for adjoint support in MulticlassLDA (#192)
• show function calls show(io,M), not show(io, MIME, M) (#194)
• error while fitting PCA model (#196)

Merged pull requests:

• Metric MDS (#180) (@wildart)
• Expend input type for MC-LDA (#193) (@wildart)
• Remove number of classes parameter from *-LDA (#199) (@wildart)

MultivariateStats v0.9.1

Diff since v0.9.0

Closed issues:

• Outdated document of PCA. (#107)
• Documentation separate from code (#128)
• Compat issue (#159)

Merged pull requests:

• Additional show info for PCA (#90) (@Tokazama)
• use API imports from StatsAPI & pairwise from StatsBase (#184) (@wildart)
• Fix what looked like a typo (#185) (@KronosTheLate)

MultivariateStats v0.9.0

Diff since v0.8.0

Closed issues:

• Less restrictive types for reconstruct (#139)
• Fitting PCA on AbstractArrays (#141)
• KPCA transform strange output (pairwise suspect) (#143)
• Multidimensional scaling (MDS) point order (#149)
• Stable docs 404 (#154)
• Bug with residual variances for PCA/PPCA (#156)
• bugs in facm() (#157)
• bugs in faem() (#158)
• ICA algorithm seems wrong (#160)
• StackoverflowError (#164)

Merged pull requests:

• Modernize struct names in docs (#135) (@mrkrause)
• Remove nonexistent export of invsqrtm (#142) (@nalimilan)
• Refactor whitening for closer integration with StatsBase types (#144) (@wildart)
• updated CI to use GH actions (#145) (@wildart)
• Lift type restrictions (#146) (@Kolaru)
• Documenter-based docs (#147) (@wildart)
• fixed pairwise for non-symmetric kernels (#148) (@wildart)
• add RegressionModel interface to LDA methods (#150) (@wildart)
• Refactor MC-LDA API and docs (#155) (@wildart)
• Refactor PCA and docs (#163) (@wildart)
• Refactor MDS code and docs (for #109) (#168) (@wildart)
• Switch to StableRNGs for tests (#169) (@kescobo)
• Refactor regression code and docs (for #109) (#170) (@wildart)
• Refactor CCA (#172) (@wildart)
• Refactor KPCA (#173) (@wildart)
• Refactor ICA (#174) (@wildart)
• Refactor PPCA (#175) (@wildart)
• Refactor FA (#176) (@wildart)
• More fixes to new docs (#177) (@wildart)
• Release new version (#178) (@wildart)

MultivariateStats v0.8.0

Diff since v0.7.0

Closed issues:

• Major release before #94? (#102)
• preprocess_mean throws TypeError when handling Integer types (#112)
• Does this library welcome implementing some robustified PCAs? (#129)
• ERROR: MethodError: no method matching PCA() (#132)

Merged pull requests:

• LREG: deprecate trans argument (#100) (@wildart)
• updated lda.rst (#108) (@OkonSamuel)
• Install TagBot as a GitHub Action (#115) (@JuliaTagBot)
• Typo (#116) (@PallHaraldsson)
• fix #112: remove type constraint (#117) (@wildart)
• Update pca.rst (#124) (@markushhh)
• loosen type bounds in MulticlassLDA and MulticlassldaStats (#127) (@OkonSamuel)
• Require distances keyword for MDS (#131) (@timholy)
• Bandaids for busted tests (#133) (@timholy)

v0.7.0 (2019-09-05)

Diff since v0.6.0

Closed issues:

• Docs link is broken. (#104)
• transform(kpca) returns error (#96)
• Change warn to at-warn for Julia 1.0 compatibility (#91)
• SubspaceLDA not working for Float32 data (#89)
• standardize data before pca (#86)
• DomainError for FactorAnalysis (#81)
• Support for custom covariance estimation methods (#80)
• The example for multivariate linear regression in the docs does not work. (#79)
• JuliaBox Package unresolved (#78)
• the PCA input X shape should be (n, d) rather than (d, n) (#77)
• Package compatibility caps (#74)
• 1.0.0 release? (#73)
• deprecation warnings are still there in v0.5.0 (#72)
• no method match fit while running PCA (#71)
• Returning eigenvalues from classical_mds? (#60)
• Implement PCoA (principal coordinate analysis) (#58)
• Make CCA take a type parameter (#53)
• Any plans for stress evaluation of MDS? (#51)

Merged pull requests:

• Bump minor release (#106) (rofinn)
• Delete link to pkg.julialang.org and move docs to the top (#105) (oxinabox)
• Remove mention of DimensionalityReduction.jl (#103) (nalimilan)
• stop using DataArray for conversion (#101) (tk3369)
• KPCA: Optimize an embedding calculation (#99) (wildart)
• KPCA: throw proper error in transform\(M\) (#97) (wildart)
• Fix subspace LDA input type errors (#93) (wildart)
• cmds: change warn to at-warn for Julia 1.0+ compatibility (#92) (nico202)
• New MDS API (#85) (wildart)
• Support for custom covariance estimation algorithms in LDA (#84) (mateuszbaran)
• Add testsets (#83) (wildart)
• Minor type & logging fixes (#82) (wildart)
• Fixed CCA type constraints (#76) (wildart)
• Various ICA fixes (#75) (wildart)

Julia v1.0 support