diff --git a/.nojekyll b/.nojekyll
index 3b6d16051..ea67f5217 100644
--- a/.nojekyll
+++ b/.nojekyll
@@ -1 +1 @@
-7956d809
\ No newline at end of file
+d788ac4f
\ No newline at end of file
diff --git a/changelog.html b/changelog.html
index 3594c16f1..91b296047 100644
--- a/changelog.html
+++ b/changelog.html
@@ -138,6 +138,9 @@ <h3 class="anchored" data-anchor-id="maintenance-and-fixes">Maintenance and fixe
 </section>
 <section id="documentation" class="level3">
 <h3 class="anchored" data-anchor-id="documentation">Documentation</h3>
+<ul>
+<li>Our Code of Conduct now includes how to send a report (#783)</li>
+</ul>
 </section>
 <section id="deprecation" class="level3">
 <h3 class="anchored" data-anchor-id="deprecation">Deprecation</h3>
diff --git a/search.json b/search.json
index 18f72fca0..6549be7cc 100644
--- a/search.json
+++ b/search.json
@@ -109,7 +109,7 @@
     "href": "changelog.html",
     "title": "Bambi",
     "section": "",
-    "text": "Add configuration facilities to Bambi (#745)\nInterpet submodule now outputs informative messages when computing default values (#745)\nBambi supports weighted responses (#761)\nBambi supports constrained responses (#764)\n\n\n\n\n\n\n\n\n\n\n\n\n\n\nThis is the first version of Bambi that is released with a Governance structure. Added in #709.\n\n\n\nBambi now supports censored responses (#697)\nImplement \"exponential\" and \"weibull\" families (#697)\nAdd \"kidney\" dataset (#697)\nAdd interpret submodule (#684, #695, #699, #701, #732, #736)\n\nImplements comparisons, predictions, slopes, plot_comparisons, plot_predictions, and plot_slopes\n\nSupport censored families\n\n\n\n\n\nBump quartodoc version to 0.6.1 (#720)\nReplace univariate_ordered with ordered (#724)\nAdd missing docstring for center_predictors (#726)\nFix bugs in plot_comparison (#731)\n\n\n\n\n\nAdd docstrings to utility functions (#696)\nMigrate documentation to Quarto (#712)\nAdd case study for MRP (#716)\nAdd example about ordinal regression (#719)\nAdd example about zero inflated models (#725)\nAdd example about predictions for new groups (#734)\n\n\n\n\n\nDrop official suport for Python 3.8 (#720)\nChange plots submodule name to interpret (#705)\n\n\n\n\n\n\n\n\nImplement new families \"ordinal\" and \"sratio\" for modeling of ordinal responses (#678)\nAllow families to implement a custom create_extra_pps_coord() (#688)\nAllow predictions on new groups (#693)\n\n\n\n\n\nRobustify how Bambi handles dims (#682)\nFix links in FAQ (#686)\nUpdate additional dependencies install command (#689)\nUpdate predict pps docstring (#690)\nAdd warning for aliases athat aren’t used (#691)\n\n\n\n\n\nAdd families to the Getting Started guide (#683)\n\n\n\n\n\n\n\n\n\n\n\nAdd support for Gaussian Processes via the HSGP approximation (#632)\nAdd new families: \"zero_inflated_poisson\", \"zero_inflated_binomial\", and \"zero_inflated_negativebinomial\" (#654)\nAdd new families: \"beta_binomial\" and \"dirichlet_multinomial\" (#659)\nAllow plot_cap() to show predictions at the observation level (#668)\nAdd new families: \"hurdle_gamma\", \"hurdle_lognormal\", \"hurdle_negativebinomial\", and \"hurdle_poisson\" (#676)\n\n\n\n\n\nModify how HSGP is built in PyMC when there are groups (#661)\nModify how Bambi is imported in the tests (#662)\nPrevent underscores from being removed in dim names (#664)\nBump sphix dependency to a version greater than 7 (#672)\n\n\n\n\n\nDocument how to use custom priors (#656)\nFix name of arviz traceplot function in the docs (#666)\nAdd example that shows how plot_cap() works (#670)\n\n\n\n\n\n\n\n\n\n\n\nRefactored the codebase to support distributional models (#607)\nAdded a default method to handle posterior predictive sampling for custom families (#625)\nplot_cap() gains a new argument target that allows to plot different parameters of the response distribution (#627)\n\n\n\n\n\nMoved the tests directory to the root of the repository (#607)\nDon’t pass dims to the response of the likelihood distribution anymore (#629)\nRemove requirements.txt and replace with pyproject.toml config file to distribute the package (#631)\n\n\n\n\n\nUpdate examples to work with the new internals (#607)\nFixed figure in the Sleepstudy example (#607)\nAdd example using distributional models (#641)\n\n\n\n\n\nRemoved versioned documentation webpage (#616)\nRemoved correlated priors for group-specific terms (#607)\nDictionary with tuple keys are not allowed for priors anymore (#607)\n\n\n\n\n\n\n\n\nUpdate to PyMC >= 5, which means we use PyTensor instead of Aesara now (#613, #614)\n\n\n\n\n\n\n\n\nImplement censored() (#581)\nAdd Formula class (#585)\nAdd common numpy transforms to extra_namespace (#589)\nAdd AsymmetricLaplace family for Quantile Regression (#591)\nAdd ‘transforms’ argument to plot_cap() (#594)\nAdd panel covariates to plot_cap() and make it more flexible (#596)\n\n\n\n\n\nReimplemented predictions to make better usage of xarray data structures (#573)\nKeep 0 dimensional parameters as 0 dimensional instead of 1 dimensional (#575)\nRefactor terms for modularity and extensibility (#582)\nRemove seed argument from model.initial_point() (#592)\nAdd build check function on prior predictive and plot prior (#605)\n\n\n\n\n\nAdd quantile regression example (#608)\n\n\n\n\n\nRemove automatic_priors argument from Model (#603)\nRemove string option for data input in Model (#604)\n\n\n\n\n\n\n\n\nAdd support for jax sampling via numpyro and blackjax samplers (#526)\nAdd Laplace family (#524)\nImprove Laplace computation and integration (#555 and #563)\n\n\n\n\n\nEnsure order variable is preserved when ploting priors (#529)\nTreat offset accordingly (#534)\nRefactor tests to share data generation code (#531)\n\n\n\n\n\nUpdate documentation following good inferencedata practices (#537)\nAdd logos to repo and docs (#542)\n\n\n\n\n\nDeprecate method argument in favor of inference_method (#554)\n\n\n\n\n\n\n\n\nBambi now uses PyMC 4.0 as it’s backend. Most if not all your previous model should run the same, without the need of any change.\nAdd Plot Conditional Adjusted Predictions plot_cap (#517)\n\n\n\n\n\nGroup specific terms now work with numeric of multiple columns (#516)\n\n\n\n\n\n\n\n\nAdd VonMises (\"vonmises\") built-in family (#453)\nModel.predict() gains a new argument include_group_specific to determine if group-specific effects are considered when making predictions (#470)\nAdd Multinomial (\"multinomial\") built-in family (#490)\n\n\n\n\n\nAdd posterior predictive sampling method to “categorical” family (#458)\nRequire Python >= 3.7.2 to fix NoReturn type bug in Python (#463)\nFixed the wrong builtin link given by link=\"inverse\" was wrong. It returned the same result as link=\"cloglog\" (#472)\nReplaced plain dictionaries with namedtuples when same dictionary structure was repeated many times (#472)\nThe function check_full_rank() in utils.py now checks the array is 2 dimensional (#472)\nRemoved _extract_family_prior() from bambi/families as it was unnecesary (#472)\nRemoved bambi/families/utils.py as it was unnecesary (#472)\nRemoved external links and unused datasets (#483)\nReplaced \"_coord_group_factor\" with \"__factor_dim\" and \"_coord_group_expr\" with \"__expr_dim\" in dimension/coord names (#499)\nFixed a bug related to modifying the types of the columns in the original data frame (#502)\n\n\n\n\n\nAdd circular regression example (#465)\nAdd Categorical regression example (#457)\nAdd Beta regression example (#442)\nAdd Radon Example (#440)\nFix typos and clear up writing in some docs (#462)\nDocumented the module bambi/defaults (#472)\nImproved documentation and made it more consistent (#472)\nCleaned Strack RRR example (#479)\n\n\n\n\n\nRemoved old default priors (#474)\nRemoved draws parameter from Model.predict() method (#504)\n\n\n\n\n\n\n\n\nFixed bug related to the shape of 2 level categorical group-specific effects (#441)\n\n\n\n\n\n\n\n\nAdd “categorical” built-in family (#426)\nAdd include_mean argument to the method Model.fit() (#434)\nAdd .set_alias() method to Model (#435)\n\n\n\n\n\nCodebase for the PyMC backend has been refactored (#408)\nFix examples that averaged posterior values across chains (#429)\nFix issue #427 with automatic priors for the intercept term (#430)\n\n\n\n\n\nAdd StudentT regression example, thanks to @tjburch (#414)\nAdd B-Spline regression example with cherry blossoms dataset (#416)\nAdd hirarchical linear regression example with sleepstudy dataset (#424)\n\n\n\n\n\n\nUse formulae 0.2.0 (#411)\n\n\n\n\n\n\n\nChange default priors for the ‘t’ family (#403)\n\n\n\n\n\nAdd installation instructions with conda (#406)\nCorrected a typo: pary_id -> party_id (#402)\nAdd donation information (#409)\n\n\n\n\n\n\n\n\nDocumentation for all versions is built from scratch when there’s a release. This ensures older versions link to the current stable release. (#396)\nAdd new axis to prior predictive samples to represent 1 chain in the InferenceData object we return (#397)\nMove Family, Likelihood and Link to the families submodule and improved some docstrings (#399)\n\n\n\n\n\nAdd example with hierarchical binomial model (#398)\n\n\n\n\n\n\n\n\nAdd alternative default priors (#360)\nAdd StudentT family (#367)\nAdd Beta family (#368)\nImplement both in-sample and out-of-sample model predictions (#372)\nAdd function to load datasets (#375)\nAdd option to specify potentials (#379)\nAdd Binomial family (#386)\n\n\n\n\n\nAutomatic switch initialization method from “jitter-adapt_diag” to “adapt_diag” when sampling fails (#383)\nPredictors are internally centered when there is an intercept. This generally results in improved sampling efficiency (#385)\nImprove documentation and error message in Model.graph() (#390)\n\n\n\n\n\nAdd Negative Binomial family example notebook (#346)\nFixed typos and improved many notebooks (#374, #377, #382)\n\n\n\n\n\n\n\n\nIt is possible to specify priors for parameters in the response distribution (#335)\nAdd probit and cloglog link functions (#340)\n\n\n\n\n\nInformative message when default priors fail because of perfect separation. Model can be fit with custom priors (#330)\nBreaking changes to the API. All the information related to the model goes in Model() instantiation now (#333)\nFix gamma family (#337)\nNon-default links are properly passed to statsmodels (#337)\nFix Wald family (#340)\nFix Negative binomial family (#340)\nAdd informative message when link function is not available for a given family (#340)\nUpdate formulae version to 0.0.10 (#348)\n\n\n\n\n\nNotebooks are updated to the new API (#336)\nAdd badges, update introduction and minor style changes in webpage (#344)\nAdd example using Gamma and Wald families (#345)\nWebpage theme has been updated to PyData theme (#347)\nAdd model evaluation to logistic regression example (#350)\n\n\n\n\n\n\n\n\nAdd option to save a figure from model.graph() by passing the name of a file. Figure format and resolution can also be set (#317)\nObjects of class Prior, Family and Model have nicer print methods (#326)\n\n\n\n\n\nAdd negative binomial family to config file, which was missing (#324)\nAdd test to check model compilation with families available (#327)\nUpdate formulae to version 0.0.9 (#329)\n\n\n\n\n\nFix gamma docstring (#328)\n\n\n\n\n\n\n\n\nUse formulae to parse model formulas (#299)\nAdd model representation (#300)\n\n\n\n\n\nRemove deprecation warning related to pm.sample returning idata (#295)\n\n\n\n\n\nAdd citation to Bambi preprint (#290)\nRemove reference to pystan (#292)\n\n\n\n\n\n\n\n\nAdd posterior predictive sampling (#250)\nAdd prior predictive sampling (#244)\nAdd gamma, negativebinomial and wald families (#207)\n\n\n\n\n\nUse pm.sample_prior_predictive function to sample and plot from prior (#238)\nFix FutureWarning: Support for multi-dimensional indexing (#242)\nUse last version of black (#245)\nfix broken link increase Python version (#227)\nAdd black style check on lint (#220)\nSome linting while re-reading library (#219)\nRemove future warning when converting the trace to InferenceData (#213)\nInclude missing files for sdist (#204)\nFixed if-else comparison that prevented HalfTStudent prior from being used (#205)\nSidestep plotting flat priors in plot_priors() (#258)\nGLM.fit_constrained in automatic priors now uses start_params = None (#265)\nCategorical Term within Model now have Term.categorical equal to True(#269)\nUse logging instead of warnings (#270)\nOmits ploting group-level effects and offset variables (#276)\nLogistic regression works with no explicit index (#277)\nAdd argument to optionally keep offsets in InferenceData (#288)\nAdd argument to optionally keep group level effects and offsets variables in plot_prior (#288)\n\n\n\n\n\nUpdate example notebooks (#232)\nadd missing notebooks (#229)\nFix notebooks (#222)\nClean docs (#200)\nAdded notebook using Bambi and ArviZ for model comparison (#267)\nUse same color palette in all notebooks (#282)\nFix divergences in examples (two divergences remaining in Strack RRR example) (#282)\n\n\n\n\n\nDrop support python 3.6 (#218)\nRemove stan backend and replace sd with sigma (#205)\nDeprecate samples argument in favor of draws (#247)\n\n\n\n\n\n\n\n\nAdd laplace approximation (#184) (only for educational use, do not use for real problems)\nUse arviz (#182, #178, #166, #159)\n\n\n\n\n\nUpdate requirements (#191)\nChange default sd prior and update docs (#189)\nAdd f-strings and support python 3.6+ (#188)\nFix parallel sampling (#186)\nLint code (#175, #173, #171, #167)\nMove coverage configuration to setup.cfg (#168)\nAdd long description to setup.py; light linting on setup.py (#162)\nBlack list external/ and tests/from pylint\n\n\n\n\n\nAdd missing example (#194)\nUpdate docs and fix typos (#185, #181)\nAdd missing items to readme and code of conduct (#180)\nSimplify readme (#179)\nUnify docstring style and remove not used code (#169)\n\n\n\n\n\nDeprecate Stan backend (#183)\n\n\n\n\n\n\n\n\nUse a callable as link function (#147)\n\n\n\n\n\nUpdate to Python 3, black and some pylint (#158)\nFix test warnings (#144)\nReorder requirements; Add matplotlib to requirements.txt (#143)\nReorder imports; Only import necessary submodules from statsmodels (#142)\nUpdate travis config (#135)\n\n\n\n\n\nAdd contributing guide (#146)\nUpdate notebooks (#140)\n\n\n\n\n\nLast version to support Python 2.7\n\n\n\n\n\n\nMinor release for bugfixes and minor improvements. Changes include:\nBug that was causing an incorrect link function to be used in the PyMC3 backend when fitting logistic models.\nFixed handling of missing values in categorical variables.\nFixed bug in set_priors() when passing numerical values for scale.\nImproved internal handling of custom priors.\nPreliminary Sphinx docs (WIP; thanks to @ejolly).\n\n\n\n\nThis is a major release that introduces several new features, significant API changes, and a large number of bug fixes and minor improvements. Notable changes include:\n\nSupport for Stan as the sampling back-end (in addition to PyMC3), via the PyStan package.\nDropped support for the add_term API; all model specification is now done via formulas.\nExpanded support for arbitrary random effects specifications; any formula now supported by patsy can be passed in as the left-hand side of a random effects specification (e.g., previously, ‘(a*b)|c’ would not have worked).\nCompletely refactored Results classes that no longer depend on PyMC3, providing a completely generic representation of sampler results, independent of any back-end.\nRefactored plotting and summary methods implemented on the abstract MCMCResults classes rather than at the back-end level.\nMuch better compilation and sampling performance for models that include random effects with many levels. In many cases, performance should now be comparable to the most efficient native implementations of the models in the respective back-ends.\nAll random effects priors now use the “non-centered” parameterization by default, significantly reducing bias for some models.\nImproved naming conventions that are more consistent with other packages (e.g., random effects now include the ‘|’ operator in term names).\nRefactored Term class, including a separate subclass for RandomTerms, and a number of other associated changes to the internal object model.\nUpdated documentation and notebooks, including two new notebooks featuring well-developed examples (datasets included).\nImproved handling of NA values in continuous columns.\nSupport for flat priors everywhere (by setting auto_scale=False).\nNumerous bug fixes and minor improvements\n\n\n\n\n\nWeakly informative default priors now work the same for all response families & link functions\nMinor bug fixes/tweaks\n\n\n\n\n\nFixes referencing of Theano ops after PyMC3 namespace clean-up\nAdded example Jupyter notebooks\nImproved handling of priors\nImproved prior plots and result summaries\nImproved access to MCMC trace results\nAdd handling for datasets with NaN values\nAdded travis-ci and coveralls support\nMinor bug fixes/tweaks\n\n\n\n\nFirst official release."
+    "text": "Add configuration facilities to Bambi (#745)\nInterpet submodule now outputs informative messages when computing default values (#745)\nBambi supports weighted responses (#761)\nBambi supports constrained responses (#764)\n\n\n\n\n\n\n\n\nOur Code of Conduct now includes how to send a report (#783)\n\n\n\n\n\n\n\n\nThis is the first version of Bambi that is released with a Governance structure. Added in #709.\n\n\n\nBambi now supports censored responses (#697)\nImplement \"exponential\" and \"weibull\" families (#697)\nAdd \"kidney\" dataset (#697)\nAdd interpret submodule (#684, #695, #699, #701, #732, #736)\n\nImplements comparisons, predictions, slopes, plot_comparisons, plot_predictions, and plot_slopes\n\nSupport censored families\n\n\n\n\n\nBump quartodoc version to 0.6.1 (#720)\nReplace univariate_ordered with ordered (#724)\nAdd missing docstring for center_predictors (#726)\nFix bugs in plot_comparison (#731)\n\n\n\n\n\nAdd docstrings to utility functions (#696)\nMigrate documentation to Quarto (#712)\nAdd case study for MRP (#716)\nAdd example about ordinal regression (#719)\nAdd example about zero inflated models (#725)\nAdd example about predictions for new groups (#734)\n\n\n\n\n\nDrop official suport for Python 3.8 (#720)\nChange plots submodule name to interpret (#705)\n\n\n\n\n\n\n\n\nImplement new families \"ordinal\" and \"sratio\" for modeling of ordinal responses (#678)\nAllow families to implement a custom create_extra_pps_coord() (#688)\nAllow predictions on new groups (#693)\n\n\n\n\n\nRobustify how Bambi handles dims (#682)\nFix links in FAQ (#686)\nUpdate additional dependencies install command (#689)\nUpdate predict pps docstring (#690)\nAdd warning for aliases athat aren’t used (#691)\n\n\n\n\n\nAdd families to the Getting Started guide (#683)\n\n\n\n\n\n\n\n\n\n\n\nAdd support for Gaussian Processes via the HSGP approximation (#632)\nAdd new families: \"zero_inflated_poisson\", \"zero_inflated_binomial\", and \"zero_inflated_negativebinomial\" (#654)\nAdd new families: \"beta_binomial\" and \"dirichlet_multinomial\" (#659)\nAllow plot_cap() to show predictions at the observation level (#668)\nAdd new families: \"hurdle_gamma\", \"hurdle_lognormal\", \"hurdle_negativebinomial\", and \"hurdle_poisson\" (#676)\n\n\n\n\n\nModify how HSGP is built in PyMC when there are groups (#661)\nModify how Bambi is imported in the tests (#662)\nPrevent underscores from being removed in dim names (#664)\nBump sphix dependency to a version greater than 7 (#672)\n\n\n\n\n\nDocument how to use custom priors (#656)\nFix name of arviz traceplot function in the docs (#666)\nAdd example that shows how plot_cap() works (#670)\n\n\n\n\n\n\n\n\n\n\n\nRefactored the codebase to support distributional models (#607)\nAdded a default method to handle posterior predictive sampling for custom families (#625)\nplot_cap() gains a new argument target that allows to plot different parameters of the response distribution (#627)\n\n\n\n\n\nMoved the tests directory to the root of the repository (#607)\nDon’t pass dims to the response of the likelihood distribution anymore (#629)\nRemove requirements.txt and replace with pyproject.toml config file to distribute the package (#631)\n\n\n\n\n\nUpdate examples to work with the new internals (#607)\nFixed figure in the Sleepstudy example (#607)\nAdd example using distributional models (#641)\n\n\n\n\n\nRemoved versioned documentation webpage (#616)\nRemoved correlated priors for group-specific terms (#607)\nDictionary with tuple keys are not allowed for priors anymore (#607)\n\n\n\n\n\n\n\n\nUpdate to PyMC >= 5, which means we use PyTensor instead of Aesara now (#613, #614)\n\n\n\n\n\n\n\n\nImplement censored() (#581)\nAdd Formula class (#585)\nAdd common numpy transforms to extra_namespace (#589)\nAdd AsymmetricLaplace family for Quantile Regression (#591)\nAdd ‘transforms’ argument to plot_cap() (#594)\nAdd panel covariates to plot_cap() and make it more flexible (#596)\n\n\n\n\n\nReimplemented predictions to make better usage of xarray data structures (#573)\nKeep 0 dimensional parameters as 0 dimensional instead of 1 dimensional (#575)\nRefactor terms for modularity and extensibility (#582)\nRemove seed argument from model.initial_point() (#592)\nAdd build check function on prior predictive and plot prior (#605)\n\n\n\n\n\nAdd quantile regression example (#608)\n\n\n\n\n\nRemove automatic_priors argument from Model (#603)\nRemove string option for data input in Model (#604)\n\n\n\n\n\n\n\n\nAdd support for jax sampling via numpyro and blackjax samplers (#526)\nAdd Laplace family (#524)\nImprove Laplace computation and integration (#555 and #563)\n\n\n\n\n\nEnsure order variable is preserved when ploting priors (#529)\nTreat offset accordingly (#534)\nRefactor tests to share data generation code (#531)\n\n\n\n\n\nUpdate documentation following good inferencedata practices (#537)\nAdd logos to repo and docs (#542)\n\n\n\n\n\nDeprecate method argument in favor of inference_method (#554)\n\n\n\n\n\n\n\n\nBambi now uses PyMC 4.0 as it’s backend. Most if not all your previous model should run the same, without the need of any change.\nAdd Plot Conditional Adjusted Predictions plot_cap (#517)\n\n\n\n\n\nGroup specific terms now work with numeric of multiple columns (#516)\n\n\n\n\n\n\n\n\nAdd VonMises (\"vonmises\") built-in family (#453)\nModel.predict() gains a new argument include_group_specific to determine if group-specific effects are considered when making predictions (#470)\nAdd Multinomial (\"multinomial\") built-in family (#490)\n\n\n\n\n\nAdd posterior predictive sampling method to “categorical” family (#458)\nRequire Python >= 3.7.2 to fix NoReturn type bug in Python (#463)\nFixed the wrong builtin link given by link=\"inverse\" was wrong. It returned the same result as link=\"cloglog\" (#472)\nReplaced plain dictionaries with namedtuples when same dictionary structure was repeated many times (#472)\nThe function check_full_rank() in utils.py now checks the array is 2 dimensional (#472)\nRemoved _extract_family_prior() from bambi/families as it was unnecesary (#472)\nRemoved bambi/families/utils.py as it was unnecesary (#472)\nRemoved external links and unused datasets (#483)\nReplaced \"_coord_group_factor\" with \"__factor_dim\" and \"_coord_group_expr\" with \"__expr_dim\" in dimension/coord names (#499)\nFixed a bug related to modifying the types of the columns in the original data frame (#502)\n\n\n\n\n\nAdd circular regression example (#465)\nAdd Categorical regression example (#457)\nAdd Beta regression example (#442)\nAdd Radon Example (#440)\nFix typos and clear up writing in some docs (#462)\nDocumented the module bambi/defaults (#472)\nImproved documentation and made it more consistent (#472)\nCleaned Strack RRR example (#479)\n\n\n\n\n\nRemoved old default priors (#474)\nRemoved draws parameter from Model.predict() method (#504)\n\n\n\n\n\n\n\n\nFixed bug related to the shape of 2 level categorical group-specific effects (#441)\n\n\n\n\n\n\n\n\nAdd “categorical” built-in family (#426)\nAdd include_mean argument to the method Model.fit() (#434)\nAdd .set_alias() method to Model (#435)\n\n\n\n\n\nCodebase for the PyMC backend has been refactored (#408)\nFix examples that averaged posterior values across chains (#429)\nFix issue #427 with automatic priors for the intercept term (#430)\n\n\n\n\n\nAdd StudentT regression example, thanks to @tjburch (#414)\nAdd B-Spline regression example with cherry blossoms dataset (#416)\nAdd hirarchical linear regression example with sleepstudy dataset (#424)\n\n\n\n\n\n\nUse formulae 0.2.0 (#411)\n\n\n\n\n\n\n\nChange default priors for the ‘t’ family (#403)\n\n\n\n\n\nAdd installation instructions with conda (#406)\nCorrected a typo: pary_id -> party_id (#402)\nAdd donation information (#409)\n\n\n\n\n\n\n\n\nDocumentation for all versions is built from scratch when there’s a release. This ensures older versions link to the current stable release. (#396)\nAdd new axis to prior predictive samples to represent 1 chain in the InferenceData object we return (#397)\nMove Family, Likelihood and Link to the families submodule and improved some docstrings (#399)\n\n\n\n\n\nAdd example with hierarchical binomial model (#398)\n\n\n\n\n\n\n\n\nAdd alternative default priors (#360)\nAdd StudentT family (#367)\nAdd Beta family (#368)\nImplement both in-sample and out-of-sample model predictions (#372)\nAdd function to load datasets (#375)\nAdd option to specify potentials (#379)\nAdd Binomial family (#386)\n\n\n\n\n\nAutomatic switch initialization method from “jitter-adapt_diag” to “adapt_diag” when sampling fails (#383)\nPredictors are internally centered when there is an intercept. This generally results in improved sampling efficiency (#385)\nImprove documentation and error message in Model.graph() (#390)\n\n\n\n\n\nAdd Negative Binomial family example notebook (#346)\nFixed typos and improved many notebooks (#374, #377, #382)\n\n\n\n\n\n\n\n\nIt is possible to specify priors for parameters in the response distribution (#335)\nAdd probit and cloglog link functions (#340)\n\n\n\n\n\nInformative message when default priors fail because of perfect separation. Model can be fit with custom priors (#330)\nBreaking changes to the API. All the information related to the model goes in Model() instantiation now (#333)\nFix gamma family (#337)\nNon-default links are properly passed to statsmodels (#337)\nFix Wald family (#340)\nFix Negative binomial family (#340)\nAdd informative message when link function is not available for a given family (#340)\nUpdate formulae version to 0.0.10 (#348)\n\n\n\n\n\nNotebooks are updated to the new API (#336)\nAdd badges, update introduction and minor style changes in webpage (#344)\nAdd example using Gamma and Wald families (#345)\nWebpage theme has been updated to PyData theme (#347)\nAdd model evaluation to logistic regression example (#350)\n\n\n\n\n\n\n\n\nAdd option to save a figure from model.graph() by passing the name of a file. Figure format and resolution can also be set (#317)\nObjects of class Prior, Family and Model have nicer print methods (#326)\n\n\n\n\n\nAdd negative binomial family to config file, which was missing (#324)\nAdd test to check model compilation with families available (#327)\nUpdate formulae to version 0.0.9 (#329)\n\n\n\n\n\nFix gamma docstring (#328)\n\n\n\n\n\n\n\n\nUse formulae to parse model formulas (#299)\nAdd model representation (#300)\n\n\n\n\n\nRemove deprecation warning related to pm.sample returning idata (#295)\n\n\n\n\n\nAdd citation to Bambi preprint (#290)\nRemove reference to pystan (#292)\n\n\n\n\n\n\n\n\nAdd posterior predictive sampling (#250)\nAdd prior predictive sampling (#244)\nAdd gamma, negativebinomial and wald families (#207)\n\n\n\n\n\nUse pm.sample_prior_predictive function to sample and plot from prior (#238)\nFix FutureWarning: Support for multi-dimensional indexing (#242)\nUse last version of black (#245)\nfix broken link increase Python version (#227)\nAdd black style check on lint (#220)\nSome linting while re-reading library (#219)\nRemove future warning when converting the trace to InferenceData (#213)\nInclude missing files for sdist (#204)\nFixed if-else comparison that prevented HalfTStudent prior from being used (#205)\nSidestep plotting flat priors in plot_priors() (#258)\nGLM.fit_constrained in automatic priors now uses start_params = None (#265)\nCategorical Term within Model now have Term.categorical equal to True(#269)\nUse logging instead of warnings (#270)\nOmits ploting group-level effects and offset variables (#276)\nLogistic regression works with no explicit index (#277)\nAdd argument to optionally keep offsets in InferenceData (#288)\nAdd argument to optionally keep group level effects and offsets variables in plot_prior (#288)\n\n\n\n\n\nUpdate example notebooks (#232)\nadd missing notebooks (#229)\nFix notebooks (#222)\nClean docs (#200)\nAdded notebook using Bambi and ArviZ for model comparison (#267)\nUse same color palette in all notebooks (#282)\nFix divergences in examples (two divergences remaining in Strack RRR example) (#282)\n\n\n\n\n\nDrop support python 3.6 (#218)\nRemove stan backend and replace sd with sigma (#205)\nDeprecate samples argument in favor of draws (#247)\n\n\n\n\n\n\n\n\nAdd laplace approximation (#184) (only for educational use, do not use for real problems)\nUse arviz (#182, #178, #166, #159)\n\n\n\n\n\nUpdate requirements (#191)\nChange default sd prior and update docs (#189)\nAdd f-strings and support python 3.6+ (#188)\nFix parallel sampling (#186)\nLint code (#175, #173, #171, #167)\nMove coverage configuration to setup.cfg (#168)\nAdd long description to setup.py; light linting on setup.py (#162)\nBlack list external/ and tests/from pylint\n\n\n\n\n\nAdd missing example (#194)\nUpdate docs and fix typos (#185, #181)\nAdd missing items to readme and code of conduct (#180)\nSimplify readme (#179)\nUnify docstring style and remove not used code (#169)\n\n\n\n\n\nDeprecate Stan backend (#183)\n\n\n\n\n\n\n\n\nUse a callable as link function (#147)\n\n\n\n\n\nUpdate to Python 3, black and some pylint (#158)\nFix test warnings (#144)\nReorder requirements; Add matplotlib to requirements.txt (#143)\nReorder imports; Only import necessary submodules from statsmodels (#142)\nUpdate travis config (#135)\n\n\n\n\n\nAdd contributing guide (#146)\nUpdate notebooks (#140)\n\n\n\n\n\nLast version to support Python 2.7\n\n\n\n\n\n\nMinor release for bugfixes and minor improvements. Changes include:\nBug that was causing an incorrect link function to be used in the PyMC3 backend when fitting logistic models.\nFixed handling of missing values in categorical variables.\nFixed bug in set_priors() when passing numerical values for scale.\nImproved internal handling of custom priors.\nPreliminary Sphinx docs (WIP; thanks to @ejolly).\n\n\n\n\nThis is a major release that introduces several new features, significant API changes, and a large number of bug fixes and minor improvements. Notable changes include:\n\nSupport for Stan as the sampling back-end (in addition to PyMC3), via the PyStan package.\nDropped support for the add_term API; all model specification is now done via formulas.\nExpanded support for arbitrary random effects specifications; any formula now supported by patsy can be passed in as the left-hand side of a random effects specification (e.g., previously, ‘(a*b)|c’ would not have worked).\nCompletely refactored Results classes that no longer depend on PyMC3, providing a completely generic representation of sampler results, independent of any back-end.\nRefactored plotting and summary methods implemented on the abstract MCMCResults classes rather than at the back-end level.\nMuch better compilation and sampling performance for models that include random effects with many levels. In many cases, performance should now be comparable to the most efficient native implementations of the models in the respective back-ends.\nAll random effects priors now use the “non-centered” parameterization by default, significantly reducing bias for some models.\nImproved naming conventions that are more consistent with other packages (e.g., random effects now include the ‘|’ operator in term names).\nRefactored Term class, including a separate subclass for RandomTerms, and a number of other associated changes to the internal object model.\nUpdated documentation and notebooks, including two new notebooks featuring well-developed examples (datasets included).\nImproved handling of NA values in continuous columns.\nSupport for flat priors everywhere (by setting auto_scale=False).\nNumerous bug fixes and minor improvements\n\n\n\n\n\nWeakly informative default priors now work the same for all response families & link functions\nMinor bug fixes/tweaks\n\n\n\n\n\nFixes referencing of Theano ops after PyMC3 namespace clean-up\nAdded example Jupyter notebooks\nImproved handling of priors\nImproved prior plots and result summaries\nImproved access to MCMC trace results\nAdd handling for datasets with NaN values\nAdded travis-ci and coveralls support\nMinor bug fixes/tweaks\n\n\n\n\nFirst official release."
   },
   {
     "objectID": "faq.html",