v0.5.1

FactorRotations v0.5.1

Diff since v0.5.0

Merged pull requests:

• CompatHelper: bump compat for Enzyme in [weakdeps] to 0.13, (keep existing compat) (#72) (@github-actions[bot])

v0.5.0

FactorRotations v0.5.0

Diff since v0.4.0

Breaking changes

• Enzyme.jl dependency moved to a package extension to improve package load times. If automatic differentiation is required for ComponentLoss or a custom rotation method it is now required to explicitly load Enzyme.jl via using Enzyme.

Merged pull requests:

• Move Enzyme.jl into extension (#71) (@alyst)

v0.4.0

FactorRotations v0.4.0

Diff since v0.3.1

Breaking changes

• Diverging rotations now print a warning instead of throwing a ConvergenceError

Merged pull requests:

• parallelize rotation with random starts (#70) (@p-gw)

v0.3.1

FactorRotations v0.3.1

Diff since v0.3.0

Merged pull requests:

• Improve criteria performance (#66) (@alyst)
• Package benchmarking (#67) (@p-gw)
• Bump version to 0.3.1 (#68) (@p-gw)

Closed issues:

• Implement Equamax (#27)
• Implement Parsimax (#28)
• Varimax did not converge for echelon-pattern matrix (#48)

v0.3.0

FactorRotations v0.3.0

Diff since v0.2.4

Breaking changes

• Custom rotation criteria must now implement in-place criterion_and_gradient! instead of criterion_and_gradient
• The epsilon parameter in Absolmin must now be striclty positive since epsilon = 0 leads to nonfinite gradients
• rotate now reflects the columns of the rotated loading matrix by default
• atol function argument for rotate is now called g_atol

New Features

• Columns of rotated loading matrices can now be reflected by calling reflect, reflect!, or using the reflect function argument for rotate (default: true)
• rotate gains a new function argument normalize (default: false) that performs Kaiser normalization before rotation and denormalizes the rotated loading matrix after rotation
• rotate gains a new function argument f_atol that sets the absolute tolerance for the comparison of minimum values with random starts
• Logging for rotate is improved: You can now set logperiod to control how often the optimization state is logged

Merged pull requests:

• Implement reflect (#53) (@p-gw)
• Fix Varimax (#54) (@alyst)
• Enhance logging (#55) (@alyst)
• Inline calculations (#58) (@alyst)
• CompatHelper: bump compat for Enzyme to 0.12, (keep existing compat) (#59) (@github-actions[bot])
• Tolerate nightly ci failures (#63) (@p-gw)
• only allow epsilon > 0 in Absolmin (#64) (@p-gw)

Closed issues:

• Reflect rotated matrix (#50)
• Fix documentation for kaiser_normalize! (#52)
• Switch to in-place criterion_and_gradient! (#57)
• Disallow epsilon <= 0 in Absolmin (#62)

v0.2.4

FactorRotations v0.2.4

Diff since v0.2.3

Merged pull requests:

• Add normalize argument to rotate (#51) (@p-gw)

Closed issues:

• Add normalize argument to rotate (#47)

v0.2.3

FactorRotations v0.2.3

Diff since v0.2.2

Merged pull requests:

• Bifactor Rotation and random starts (#45) (@p-gw)

Closed issues:

• Implement bifactor rotation (#44)
• Add random starts (#46)

v0.2.2

FactorRotations v0.2.2

Diff since v0.2.1

Merged pull requests:

• Performance optimisation in Varimax rotation (#43) (@p-gw)

Closed issues:

• Performance optimisation in Varimax rotation (#42)

v0.2.1

FactorRotations v0.2.1

Diff since v0.2.0

Merged pull requests:

• Implement global logging setting (#41) (@p-gw)

Closed issues:

• Set verbosity globally (#40)

v0.2.0

FactorRotations v0.2.0

Diff since v0.1.4

Merged pull requests:

• Add MultivariateStats.jl Extension (#38) (@p-gw)
• Implement custom result type (#39) (@p-gw)

Closed issues:

• Add MultivariateStats.jl support (#18)
• Return factor correlation matrix (#34)