- loo-pit plot. The kde is computed over the data interval (this could be shorter than [0, 1]). The HDI is computed analitically (#1215)
- Added
html_repr
of InferenceData objects for jupyter notebooks. (#1217) - Added support for PyJAGS via the function
from_pyjags
. (#1219 and #1245) from_pymc3
can now retrievecoords
anddims
from model context (#1228, #1240 and #1249)plot_trace
now supports multiple aesthetics to identify chain and variable shape and support matplotlib aliases (#1253)plot_hdi
can now take already computed HDI values (#1241)
- Include data from
MultiObservedRV
toobserved_data
when usingfrom_pymc3
(#1098) - Added a note on
plot_pair
when trying to useplot_kde
onInferenceData
objects. (#1218) - Added
log_likelihood
argument tofrom_pyro
and a warning if log likelihood cannot be obtained (#1227) - Skip tests on matplotlib animations if ffmpeg is not installed (#1227)
- Fix hpd bug where arguments were being ignored (#1236)
- Remove false positive warning in
plot_hdi
and fixed matplotlib axes generation (#1241) - Change the default
zorder
of scatter points from0
to0.6
inplot_pair
(#1246) - Update
get_bins
for numpy 1.19 compatibility (#1256) - Fixes to
rug
,divergences
arguments inplot_trace
(#1253)
- Using
from_pymc3
without a model context available now raises aFutureWarning
and will be deprecated in a future version (#1227) - In
plot_trace
,chain_prop
andcompact_prop
as tuples will now raise aFutureWarning
(#1253) hdi
with 2d data raises a FutureWarning (#1241)
- A section has been added to the documentation at InferenceDataCookbook.ipynb illustrating the use of ArviZ in conjunction with PyJAGS. (#1219 and #1245)
- Fixed inconsistent capitalization in
plot_hdi
docstring (#1221) - Fixed and extended
InferenceData.map
docs (#1255)
- Restructured internals of
from_pymc3
to handle old pymc3 releases and sliced traces and to provide useful warnings (#1211)
- Fixed bug in
from_pymc3
for slicedpymc3.MultiTrace
input (#1209)
- Fixed bug in
from_pymc3
when used with PyMC3<3.9 (#1203) - Fixed enforcement of rcParam
plot.max_subplots
inplot_trace
andplot_pair
(#1205) - Removed extra subplot row and column in in
plot_pair
withmarginal=True
(#1205) - Added latest PyMC3 release to CI in addition to using GitHub master (#1207)
- Use
dev
as version indicator in online documentation (#1204)
- Stats and plotting functions that provide
var_names
arg can now filter parameters based on partial naming (filter="like"
) or regular expressions (filter="regex"
) (see #1154). - Add
true_values
argument forplot_pair
. It allows for a scatter plot showing the true values of the variables #1140 - Allow xarray.Dataarray input for plots.(#1120)
- Revamped the
hpd
function to make it work with mutidimensional arrays, InferenceData and xarray objects (#1117) - Skip test for optional/extra dependencies when not installed (#1113)
- Add option to display rank plots instead of trace (#1134)
- Add out-of-sample groups (
predictions
andpredictions_constant_data
) tofrom_dict
(#1125) - Add out-of-sample groups (
predictions
andpredictions_constant_data
) andconstant_data
group to pyro and numpyro translation (#1090, #1125) - Add
num_chains
andpred_dims
arguments to from_pyro and from_numpyro (#1090, #1125) - Integrate jointplot into pairplot, add point-estimate and overlay of plot kinds (#1079)
- New grayscale style. This also add two new cmaps
cet_grey_r
andcet_grey_r
. These are perceptually uniform gray scale cmaps from colorcet (linear_grey_10_95_c0) (#1164) - Add warmup groups to InferenceData objects, initial support for PyStan (#1126) and PyMC3 (#1171)
hdi_prob
will not plot hdi if argumenthide
is passed. Previouslycredible_interval
would omit HPD ifNone
was passed (#1176)- Add
stats.ic_pointwise
rcParam (#1173) - Add
var_name
argument to information criterion calculation:compare
,loo
andwaic
(#1173)
- Fixed
plot_pair
functionality for two variables with bokeh backend (#1179) - Changed
diagonal
argument formarginals
and fixedpoint_estimate_marker_kwargs
inplot_pair
(#1167) - Fixed behaviour of
credible_interval=None
inplot_posterior
(#1115) - Fixed hist kind of
plot_dist
with multidimensional input (#1115) - Fixed
TypeError
intransform
argument ofplot_density
andplot_forest
whenInferenceData
is a list or tuple (#1121) - Fixed overlaid pairplots issue (#1135)
- Update Docker building steps (#1127)
- Updated benchmarks and moved to asv_benchmarks/benchmarks (#1142)
- Moved
_fast_kde
,_fast_kde_2d
,get_bins
and_sturges_formula
tonumeric_utils
andget_coords
toutils
(#1142) - Rank plot: rename
axes
argument toax
(#1144) - Added a warning specifying log scale is now the default in compare/loo/waic functions (#1150).
- Fixed bug in
plot_posterior
with rcParam "plot.matplotlib.show" = True (#1151) - Set
fill_last
argument ofplot_kde
to False by default (#1158) - plot_ppc animation: improve docs and error handling (#1162)
- Fix import error when wrapped function docstring is empty (#1192)
- Fix passing axes to plot_density with several datasets (#1198)
hpd
function deprecated in favor ofhdi
.credible_interval
argument replaced byhdi_prob
throughout with exception ofplot_loo_pit
(#1176)plot_hpd
function deprecated in favor ofplot_hdi
. (#1190)
- Add classifier to
setup.py
including Matplotlib framework (#1133) - Image thumbs generation updated to be Bokeh 2 compatible (#1116)
- Add new examples for
plot_pair
(#1110) - Add examples for
psislw
andr2_score
(#1129) - Add more examples on 2D kde customization (#1158)
- Make docs compatible with sphinx3 and configure
intersphinx
for better references (#1184) - Extend the developer guide and add it to the website (#1184)
- Add out-of-sample predictions (
predictions
andpredictions_constant_data
groups) to pymc3, pystan, cmdstan and cmdstanpy translations (#983, #1032 and #1064) - Started adding pointwise log likelihood storage support (#794, #1044 and #1064)
- Add out-of-sample predictions (
predictions
andpredictions_constant_data
groups) to pymc3 and pystan translations (#983 and #1032) - Started adding pointwise log likelihood storage support (#794, #1044)
- Violinplot: rug-plot option (#997)
- Integrated rcParams
plot.point_estimate
(#994),stats.ic_scale
(#993) andstats.credible_interval
(#1017) - Added
group
argument toplot_ppc
(#1008),plot_pair
(#1009) andplot_joint
(#1012) - Added
transform
argument toplot_trace
,plot_forest
,plot_pair
,plot_posterior
,plot_rank
,plot_parallel
,plot_violin
,plot_density
,plot_joint
(#1036) - Add
skipna
argument tohpd
andsummary
(#1035) - Added
transform
argument toplot_trace
,plot_forest
,plot_pair
,plot_posterior
,plot_rank
,plot_parallel
,plot_violin
,plot_density
,plot_joint
(#1036) - Add
marker
functionality tobokeh_plot_elpd
(#1040) - Add
ridgeplot_quantiles
argument toplot_forest
(#1047) - Added the functionality interactive legends for bokeh plots of
densityplot
,energyplot
andessplot
(#1024) - New defaults for cross validation:
loo
(old: waic) andlog
-scale (old:deviance
-scale) (#1067) - Experimental Feature: Added
arviz.wrappers
module to allow ArviZ to refit the models if necessary (#771) - Experimental Feature: Added
reloo
function to ArviZ (#771) - Added new helper function
matplotlib_kwarg_dealiaser
(#1073) - ArviZ version to InferenceData attributes. (#1086)
- Add
log_likelihood
argument tofrom_pymc3
(#1082) - Integrated rcParams for
plot.bokeh.layout
andplot.backend
. (#1089) - Add automatic legends in
plot_trace
with compact=True (matplotlib only) (#1070) - Updated hover information for
plot_pair
with bokeh backend (#1074)
- Fixed bug in density and posterior plot bin computation (#1049)
- Fixed bug in density plot ax argument (#1049)
- Fixed bug in extracting prior samples for cmdstanpy (#979)
- Fix erroneous warning in traceplot (#989)
- Correct bfmi denominator (#991)
- Removed parallel from jit full (#996)
- Rename flat_inference_data_to_dict (#1003)
- Violinplot: fix histogram (#997)
- Convert all instances of SyntaxWarning to UserWarning (#1016)
- Fix
point_estimate
inplot_posterior
(#1038) - Fix interpolation
hpd_plot
(#1039) - Fix
io_pymc3.py
to handle models withpotentials
(#1043) - Fix several inconsistencies between schema and
from_pymc3
implementation in groupsprior
,prior_predictive
andobserved_data
(#1045) - Stabilize covariance matrix for
plot_kde_2d
(#1075) - Removed extra dim in
prior
data infrom_pyro
(#1071) - Moved CI and docs (build & deploy) to Azure Pipelines and started using codecov (#1080)
- Fixed bug in densityplot when variables differ between models (#1096)
from_pymc3
now requires PyMC3>=3.8
- Updated
InferenceData
schema specification (log_likelihood
,predictions
andpredictions_constant_data
groups) - Clarify the usage of
plot_joint
(#1001) - Added the API link of function to examples (#1013)
- Updated PyStan_schema_example to include example of out-of-sample prediction (#1032)
- Added example for
concat
method (#1037)
- Update for pair_plot divergences can be selected
- Default tools follow global (ArviZ) defaults
- Add interactive legend for a plot, if only two variables are used in pairplot
- Change
packaging
import from absolute to relative format, explicitly importingversion
function
- Initial bokeh support.
- ArviZ.jl a Julia interface to ArviZ (@sethaxen )
- Fully support
numpyro
(@fehiepsi ) - log_likelihood and observed data from pyro
- improve rcparams
- fix
az.concat
functionality (@anzelpwj )
- distplot docstring plotting example (@jscarbor )
- Comment dev requirements in setup.py
- Add from_numpyro Integration (#811)
- Numba Google Summer of Code additions (https://ban-zee.github.io/jekyll/update/2019/08/19/Submission.html)
- Model checking, Inference Data, and Convergence assessments (https://github.com/OriolAbril/gsoc2019/blob/master/final_work_submission.md)
- Reorder stats columns (#695)
- Plot Forest reports ess and rhat by default(#685)
- Add pointwise elpd (#678)
- Fix io_pymc3 bug (#693)
- Fix io_pymc3 warning (#686)
- Fix 0 size bug with pystan (#677)
- Add plot_dist (#592)
- New rhat and ess (#623)
- Add plot_hpd (#611)
- Add plot_rank (#625)
- Remove load_data and save_data (#625)
- Plot ppc supports multiple chains (#526)
- Plot titles now wrap (#441)
- plot_density uses a grid (#379)
- emcee reader support (#550)
- Animations in plot_ppc (#546)
- Optional dictionary for stat_funcs in summary (#583)
- Can exclude variables in selections with negated variable names (#574)
- Order maintained with xarray_var_iter (#557)
- Testing very improved (multiple)
- Fix nan handling in effective sample size (#573)
- Fix kde scaling (#582)
- xticks for discrete variables (#586)
- Empty InferenceData saves consistent with netcdf (#577)
- Removes numpy pinning (#594)
- JOSS and Zenodo badges (#537)
- Gitter badge (#548)
- Docs for combining InferenceData (#590)
- Support PyStan3 (#464)
- Add some more information to the inference data of tfp (#447)
- Use Split R-hat (#477)
- Normalize from_xyz functions (#490)
- KDE: Display quantiles (#479)
- Add multiple rope support to
plot_forest
(#448) - Numba jit compilation to speed up some methods (#515)
- Add
from_dict
for easier creation of az.InferenceData objects (#524) - Add stable logsumexp (#522)
- Fix for
from_pyro
with multiple chains (#463) - Check
__version__
for attr (#466) - And exception to plot compare (#461)
- Add Docker Testing to travisCI (#473)
- fix jointplot warning (#478)
- Fix tensorflow import bug (#489)
- Rename N_effective to S_effective (#505)
- Add docs to plot compare (#461)
- Add InferenceData tutorial in header (#502)
- Added figure to InferenceData tutorial (#510)
- Fix installation problem with release 0.3.0
- First Beta Release