performance 0.7.2 Unreleased

New functions

performance 0.7.1 2021-04-09

New supported models

  • Add support for model_fit (tidymodels).

  • model_performance supports kmeans models.

General

  • Give more informative warning when r2_bayes() for BFBayesFactor objects can’t be calculated.

  • Several check_*() functions now return informative messages for invalid model types as input.

  • r2() supports mhurdle (mhurdle) models.

  • Added print() methods for more classes of r2().

  • The performance_roc() and performance_accuracy() functions unfortunately had spelling mistakes in the output columns: Sensitivity was called Sensivity and Specificity was called Specifity. We think these are understandable mistakes :-)

Changes to functions

check_model()

model_performance()

Other functions

  • The default-method for r2() now tries to compute an r-squared for all models that have no specific r2()-method yet, by using following formula: 1-sum((y-y_hat)^2)/sum((y-y_bar)^2))

  • The column name Parameter in check_collinearity() is now more appropriately named Term.

Bug fixes

  • test_likelihoodratio() now correctly sorts models with identical fixed effects part, but different other model parts (like zero-inflation).

  • Fixed incorrect computation of models from inverse-Gaussian families, or Gaussian families fitted with glm().

  • Fixed issue in performance_roc() for models where outcome was not 0/1 coded.

  • Fixed issue in performance_accuracy() for logistic regression models when method = "boot".

  • cronbachs_alpha() did not work for matrix-objects, as stated in the docs. It now does.

performance 0.7.0 2021-02-03

General

  • Roll-back R dependency to R >= 3.4.

Breaking Changes

New functions to test or compare models

Changes to functions

performance_lrt()

  • performance_lrt() get an alias test_likelihoodratio().

  • Does not return AIC/BIC now (as they are not related to LRT per se and can be easily obtained with other functions).

  • Now contains a column with the difference in degrees of freedom between models.

  • Fixed column names for consistency.

model_performance()

  • Added more diagnostics to models of class ivreg.

Other functions

Bug fixes

performance 0.6.1 2020-12-09

General

Bug fixes

  • Fixed examples, now using suggested packages only conditionally.

performance 0.6.0 2020-12-01

General

New functions

Changes to functions

model_performance()

  • model_performance() is now more robust and doesn’t fail if an index could not be computed. Instead, it returns all indices that were possible to calculate.

  • model_performance() gains a default-method that catches all model objects not previously supported. If model object is also not supported by the default-method, a warning is given.

  • model_performance() for metafor-models now includes the degrees of freedom for Cochran’s Q.

Other functions

  • performance_mse() and performance_rmse() now always try to return the (R)MSE on the response scale.

  • performance_accuracy() now accepts all types of linear or logistic regression models, even if these are not of class lm or glm.

  • performance_roc() now accepts all types of logistic regression models, even if these are not of class glm.

  • r2() for mixed models and r2_nakagawa() gain a tolerance-argument, to set the tolerance level for singularity checks when computing random effect variances for the conditional r-squared.

Bug fixes

  • Fixed issue in icc() introduced in the last update that make lme-models fail.

  • Fixed issue in performance_roc() for models with factors as response.

performance 0.5.1 2020-10-29

Breaking changes

  • Column names for model_performance() and compare_performance() were changed to be in line with the easystats naming convention: LOGLOSS is now Log_loss, SCORE_LOG is Score_log and SCORE_SPHERICAL is now Score_spherical.

New functions

  • r2_posterior() for Bayesian models to obtain posterior distributions of R-squared.

Changes to functions

Bug fixes

  • Fix bug in compare_performance() when ... arguments were function calls to regression objects, instead of direct function calls.

performance 0.5.0 2020-09-12

General

New functions

  • pp_check() to compute posterior predictive checks for frequentist models.

Bug fixes

performance 0.4.8 2020-07-27

General

  • Removed suggested packages that have been removed from CRAN.

Changes to functions

  • icc() now also computes a “classical” ICC for brmsfit models. The former way of calculating an “ICC” for brmsfit models is now available as new function called variance_decomposition().

Bug fixes

performance 0.4.7 2020-06-14

General

  • Support for models from package mfx.

Changes to functions

  • model_performance.rma() now includes results from heterogeneity test for meta-analysis objects.

  • check_normality() now also works for mixed models (with the limitation that studentized residuals are used).

  • check_normality() gets an effects-argument for mixed models, to check random effects for normality.

Bug fixes

performance 0.4.6 2020-05-03

General

  • Minor revisions to model_performance() to meet changes in mlogit package.

  • Support for bayesx models.

Changes to functions

  • icc() gains a by_group argument, to compute ICCs per different group factors in mixed models with multiple levels or cross-classified design.

  • r2_nakagawa() gains a by_group argument, to compute explained variance at different levels (following the variance-reduction approach by Hox 2010).

  • performance_lrt() now works on lavaan objects.

Bug fixes

  • Fix issues in some functions for models with logical dependent variable.

  • Fix bug in check_itemscale(), which caused multiple computations of skewness statistics.

  • Fix issues in r2() for gam models.

performance 0.4.5 2020-03-28

General

Changes to functions

  • compare_performance() gets a bayesfactor argument, to include or exclude the Bayes factor for model comparisons in the output.

  • Added r2.aov().

Bug fixes

performance 0.4.4 2020-02-10

General

  • Removed logLik.felm(), because this method is now implemented in the lfe package.

  • Support for DirichletRegModel models.

New functions

Bug fixes

performance 0.4.3 2020-01-22

General

  • Support for mixor, cpglm and cpglmm models.

New functions

  • performance_aic() as a small wrapper that returns the AIC. It is a generic function that also works for some models that don’t have a AIC method (like Tweedie models).

  • performance_lrt() as a small wrapper around anova() to perform a Likelihood-Ratio-Test for model comparison.

Bug fixes

  • Fix issues with CRAN checks.

Changes to functions

performance 0.4.2 2019-12-11

General

Changes to functions

Minor changes

Bug fixes

performance 0.4.0 2019-10-21

General

Deprecated and Defunct

Changes to functions

Bug fixes

  • Fixed bug in compare_performance() that toggled a warning although models were fit from same data.

  • Fixed bug in check_model() for glmmTMB models that occurred when checking for outliers.

performance 0.3.0 2019-08-05

General

Breaking changes

  • The attribute for the standard error of the Bayesian R2 (r2_bayes()) was renamed from std.error to SE to be in line with the naming convention of other easystats-packages.

  • compare_performance() now shows the Bayes factor when all compared models are fit from the same data. Previous behaviour was that the BF was shown when models were of same class.

Changes to functions

  • model_performance() now also works for lavaan-objects.

  • check_outliers() gets a method-argument to choose the method for detecting outliers. Furthermore, two new methods (Mahalanobis Distance and Invariant Coordinate Selection) were implemented.

  • check_model() now performs more checks for GLM(M)s and other model objects.

  • check_model() gets a check-argument to plot selected checks only.

  • r2_nakagawa() now returns r-squared for models with singular fit, where no random effect variances could be computed. The r-squared then does not take random effect variances into account. This behaviour was changed to be in line with MuMIn::r.squaredGLMM(), which returned a value for models with singular fit.

  • check_distribution() now detects negative binomial and zero-inflated distributions. Furthermore, attempt to improve accuracy.

  • check_distribution() now also accepts a numeric vector as input.

  • compare_performance() warns if models were not fit from same data.

New check-functions

Bug fixes

  • Fixed issues with compare_performance() and row-ordering.

  • Fixed issue in check_collinearity() for zero-inflated models, where the zero-inflation component had not enough model terms to calculate multicollinearity.

  • Fixed issue in some check_*() and performance_*() functions for models with binary outcome, when outcome variable was a factor.

performance 0.2.0 2019-06-04

General

  • r2() now works for more regression models.

  • r2_bayes() now works for multivariate response models.

  • model_performance() now works for more regression models, and also includes the log-loss, proper scoring rules and percentage of correct predictions as new metric for models with binary outcome.

New performance-functions

New check-functions

New indices-functions

  • r2_mckelvey() to compute McKelvey and Zavoinas R2 value.

  • r2_zeroinflated() to compute R2 for zero-inflated (non-mixed) models.

  • r2_xu() as a crude R2 measure for linear (mixed) models.

Breaking changes

  • model_performance.stanreg() and model_performance.brmsfit() now only return one R2-value and its standard error, instead of different (robust) R2 measures and credible intervals.

  • error_rate() is now integrated in the performance_pcp()-function.

Changes to functions

  • model_performance.stanreg() and model_performance.brmsfit() now also return the WAIC (widely applicable information criterion).

  • r2_nakagawa() now calculates the full R2 for mixed models with zero-inflation.

  • icc() now returns NULL and no longer stops when no mixed model is provided.

  • compare_performance() now shows the Bayes factor when all compared models are of same class.

  • Some functions get a verbose-argument to show or suppress warnings.

Bug fixes