RobustModels v0.5.0

Diff since v0.4.5

Closed issues:

• StackOverflowError if inputs allow Missing type (#16)
• Hung Process (#32)

Merged pull requests:

• Fix missing type leading to StackOverflow (#17) (@svilupp)
• add test univariate: mean_and_std = (mean, std) (#18) (@getzze)
• add influence function expression in the docs of each estimator (#19) (@getzze)
• create an abstract type AbstractMEstimator to differentiate from non-… (#20) (@getzze)
• prefix new methods by the module defining the function to be extended… (#21) (@getzze)
• add two Catoni loss functions for M-Estimators, CatoniWideLoss and CatoniNarrowLoss (#22) (@getzze)
• add wobs (#23) (@getzze)
• bump compat StatsModels to "0.6, 0.7" (#24) (@getzze)
• remove use of nobs (#25) (@getzze)
• Correct leverage, dof for RidgePred (#26) (@getzze)
• correct Ridge lambda factor 2 and add tests (#27) (@getzze)
• Correct weights, add tests for weights (#28) (@getzze)
• Clean estimator (#30) (@getzze)
• Remove TableRegressionModel wrapper (#31) (@getzze)
• Correct infinite loop (#33) (@getzze)
• Compatibility with julia 1.3 (#34) (@getzze)
• Add GLM.DensePredQR (#35) (@getzze)
• Bump to v0.5 (#36) (@getzze)

RobustModels v0.4.5

Diff since v0.4.4

• Update dependencies compat versions (Roots)

RobustModels v0.4.4

Diff since v0.4.3

• Export hasintercept function
• Correct nulldeviance and nullloglikelihood for models without intercept (https://github.com/JuliaStats/StatsAPI.jl/pull/14).
• Update dependencies compat versions (Tulip)

Closed issues:

• Register package (#13)

Merged pull requests:

• CompatHelper: bump compat for Tulip to 0.9, (keep existing compat) (#15) (@github-actions[bot])

RobustModels v0.4.3

Diff since v0.4.2

RobustModels v0.4.2

• Minimal compatibility set to julia 1.3 (because of Tulip.jl>=0.8)

RobustModels v0.4.1

• Correctly handle multidimensional arrays with univariate robust functions.
• Reformat code.

RobustModels v0.4

• Drop the heavy JuMP dependency and use Tulip with the unstable internal API instead for quantile regression.
• Add univariate robust functions: mean, std, var, sem, mean_and_std, mean_and_var and mean_and_sem.
• Drop Coveralls support.

RobustModels v0.3

• API change: estimator type and loss functions are now two different types. Fit a robust model using the new API: rlm(form, data, MMEstimator{TukeyLoss}(); initial_scale=:L1)
• Robust Ridge regression by specifying the keyword argument ridgeλ to rlm: rlm(form, data, MEstimator{HuberLoss}(); initial_scale=:mad, ridgeλ=0.1)
• Add documentation.