profile
viewpoint
Osvaldo Martin aloctavodia IMASL-CONICET San Luis, Argentina https://aloctavodia.github.io/

aloctavodia/Doing_bayesian_data_analysis 768

Python/PyMC3 versions of the programs described in Doing bayesian data analysis by John K. Kruschke

aloctavodia/Statistical-Rethinking-with-Python-and-PyMC3 700

Python/PyMC3 port of the examples in " Statistical Rethinking A Bayesian Course with Examples in R and Stan" by Richard McElreath

aloctavodia/BAP 339

Bayesian Analysis with Python (Second Edition)

aloctavodia/Modelado_Bayesiano 15

Introducción al modelado, inferencia y análisis de modelos Bayesianos

aloctavodia/Bayesian-Analysis-with-Python 6

Bayesian Analysis with Python by Packt

aloctavodia/intro_ciencia_de_datos 6

Curso introducctorio a ciencia de datos y Python

aloctavodia/MetNum 2

Algunos Métodos Numéricos explicados usando Python

aloctavodia/density_estimation 1

This repository contains notebooks with different probability density function estimators.

aloctavodia/pymc3 1

Probabilistic Programming in Python. Uses Theano as a backend, supports NUTS and ADVI.

ahartikainen/quantile_kde 0

Quantile based fast kde method

push eventaloctavodia/Theano-PyMC

Oriol Abril-Pla

commit sha 19a336f4b31b23bcbfbc2b28c3272f7b690a0047

add codecov (#119) * add codecov * fix coverage report generation by pytest

view details

Brandon T. Willard

commit sha ae2919bf8d7a28c6f5e48041e0fb0488b77bd68d

Revert "add codecov (#119)" This reverts commit 19a336f4b31b23bcbfbc2b28c3272f7b690a0047.

view details

Abdalaziz Rashid

commit sha 6150681097433cb477a778c009180dab35e02d3b

Implement exponentiation by squaring for matrix_power Closes #117.

view details

Brandon T. Willard

commit sha 4366da4ffaec495457f18d0e4746bf8bb02d20fa

Update flake8 settings

view details

Brandon T. Willard

commit sha 0a499b81fa1d4797d39c3e56860d74c052228aaa

Add black compatible pylint settings to setup.cfg

view details

Brandon T. Willard

commit sha b70039a80b16214aa6d7a5d43fb5665292729a0b

Fix bugbears caught by flake8-bugbear and related test updates Most of the bugbears involved `assert False` and bad function keyword defaults. In tests, the former often signaled needed `pytest.raises` rewrites.

view details

Brandon T. Willard

commit sha b079be4cb8d57c4dc85ed685644207bcbadd7301

Exclude certain files from pre-commit hooks

view details

Brandon T. Willard

commit sha 51a465b17bd4f07b7f351d6425910d534bc6b350

Update Makefile target names and output messages

view details

Brandon T. Willard

commit sha 7b02965d6ded5c19d939608497356b7a57972a66

Only ignore some flake8 errors in __init__.py files

view details

Brandon T. Willard

commit sha f641f32df38fd4f0fdba01e09522c6cee8172f3c

Add isort to settings, requirements, and Makefile

view details

Brandon T. Willard

commit sha 1be3fe3b1603bdaf1e425a243186aef86ae6d89a

Apply isort to setup.py

view details

Brandon T. Willard

commit sha c4d699c72bcc502da0bf965835317096670e8096

Apply isort to top-level tests package modules

view details

Brandon T. Willard

commit sha 6f838fe0be9beb1eb9b70388834a9f9f3381337e

Apply isort to tests.compile sub-package

view details

Brandon T. Willard

commit sha ed04fa8a4a2d325c44f11d01db36ec5e84a7f0ef

Apply isort to tests.d3viz sub-package modules

view details

Brandon T. Willard

commit sha b4fbaa2ea2b2e2cc9dcb40a4a198ee16dad69144

Apply isort to tests.gof sub-package modules

view details

Brandon T. Willard

commit sha 88599bc3977c0cd20a569907553541443f24980c

Apply isort to theano.gpuarray and remove circular references Two new modules were made in order to remove the circular references: `theano.gpuarray.dnn_opt` (i.e. optimizations specific to `theano.gpuarray.dnn`) and `theano.gpuarray.optdb` (i.e. optimization databases for gpuarray `Op`s).

view details

Brandon T. Willard

commit sha a3dbdec363e1cc7079db10069ea6872d72428c56

Apply isort to tests.gpuarray sub-package modules

view details

Brandon T. Willard

commit sha 558d455d71b0267f915150b5ab296af75f5f3579

Apply isort to tests.misc sub-package modules

view details

Brandon T. Willard

commit sha e97cae435defa2632ea6384baad946826d84bf5f

Apply isort to tests.sandbox sub-package modules

view details

Brandon T. Willard

commit sha 11e773dfb5b2d948ee6f3a40f7dc9b549d6abc81

Apply isort to tests.scalar sub-package modules

view details

push time in a day

PR opened bambinos/bambi

add github ribbon

fix #246

+5 -1

0 comment

1 changed file

pr created time in a day

create barnchaloctavodia/bambi

branch : fork_ribbon

created branch time in a day

push eventaloctavodia/bambi

Osvaldo Martin

commit sha 3d0eebbbea4ddbaa8ce414a30049c9fcfe955e79

add ploting arguments (#255)

view details

Tomás Capretto

commit sha 40209f95f4836b3de867cae4d8f8a1733083b13b

Update CHANGELOG.md Listed a fix already committed but not included here.

view details

push time in a day

Pull request review commentpymc-devs/pymc3

Add Bayesian Additive Regression Trees (BARTs)

+#   Copyright 2020 The PyMC Developers+#+#   Licensed under the Apache License, Version 2.0 (the "License");+#   you may not use this file except in compliance with the License.+#   You may obtain a copy of the License at+#+#       http://www.apache.org/licenses/LICENSE-2.0+#+#   Unless required by applicable law or agreed to in writing, software+#   distributed under the License is distributed on an "AS IS" BASIS,+#   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.+#   See the License for the specific language governing permissions and+#   limitations under the License.++import numpy as np+from theano import function as theano_function++from .arraystep import ArrayStepShared, Competence+from ..distributions import BART+from ..distributions.tree import Tree+from ..model import modelcontext+from ..theanof import inputvars, make_shared_replacements, join_nonshared_inputs+++class PGBART(ArrayStepShared):+    """+    Particle Gibss BART sampling step++    Parameters+    ----------+    vars: list+        List of variables for sampler+    num_particles : int+        Number of particles for the SMC sampler. Defaults to 10

I tried to follow the nomenclature in the papers and one step of the PGBART is "conditional-SMC" method. But I see how this can be confusing.

aloctavodia

comment created time in 2 days

PullRequestReviewEvent

push eventaloctavodia/pymc3

aloctavodia

commit sha 9050469110b9ce00548cabbaf4c0a16e1bdca13f

remove space. Co-authored-by: aloctavodia <aloctavodia@gmail.com> Co-authored-by: jmloyola <jmloyola@outlook.com>

view details

push time in 2 days

push eventarviz-devs/arviz

Agustina Arroyuelo

commit sha 475e34c1bc79e5eef58ef8a12819bab18342bad8

Fix labels and use circular KDE in plot_trace (#1428) * fix circular traceplot labels * update changelog * fix xticklabels * update test

view details

push time in 2 days

PR merged arviz-devs/arviz

Fix labels and use circular KDE in plot_trace

Description

I noticed that the KDE plot in plot_trace was not using the circular KDE function (#1284). I also fixed the labels so that both columns share the same units.

Degrees:

traceplot_circ_degrees

Radians:

traceplot_circ_radians

Checklist

  • [X] Follows official PR format
  • [X] Includes a sample plot to visually illustrate the changes (only for plot-related functions)
  • [x] Code style correct (follows pylint and black guidelines)
  • [x] Changes are listed in changelog

<!-- Also, please consider reading the contributing guidelines and code of conduct carefully before submitting the PR. They are available at

  • https://github.com/arviz-devs/arviz/blob/master/CONTRIBUTING.md
  • https://github.com/arviz-devs/arviz/blob/master/CODE_OF_CONDUCT.md

We understand that PRs can sometimes be overwhelming, especially as the reviews start coming in. Please let us know if the reviews are unclear or the recommended next step seems overly demanding, if you would like help in addressing a reviewer's comments, or if you have been waiting too long to hear back on your PR. -->

+33 -18

1 comment

4 changed files

agustinaarroyuelo

pr closed time in 2 days

delete branch arviz-devs/arviz

delete branch : plot_rank_kwargs

delete time in 2 days

push eventarviz-devs/arviz

Osvaldo Martin

commit sha 9997735c2c917471e39de9bc1f0f3acf0e551054

plot_rank: add ref_line, bar, vlines and marker_vlines kwargs (#1419) * add ref_line, bar, vlines and marker_vlines kwargs * black * add example * update changelog

view details

push time in 2 days

PR merged arviz-devs/arviz

plot_rank: add ref_line, bar, vlines and marker_vlines kwargs

Fix #1396

  • [x] Follows official PR format
  • [x] Includes a sample plot to visually illustrate the changes (only for plot-related functions)
  • [x] New features are properly documented (with an example if appropriate)?
  • [x] Includes new or updated tests to cover the new feature
  • [x] Code style correct (follows pylint and black guidelines)
  • [x] Changes are listed in changelog

idata = az.load_arviz_data("centered_eight") az.plot_rank(idata, var_names="mu", kind="vlines", vlines_kwargs={'lw':0}, marker_vlines_kwargs={'lw':3}); index

az.plot_rank(idata, var_names="mu", ref_line_kwargs={"line_width":2, "line_color":"gray"}, bar_kwargs={"width":50}, backend="bokeh"); bokeh_plot

+113 -22

1 comment

6 changed files

aloctavodia

pr closed time in 2 days

issue closedarviz-devs/arviz

plot_rank ignores backend_kwargs

Using plot_rank on many a trace with many coefficients can require changing the line widths and marker sizes (e.g. when using kind="vlines").

This usage however has no effect:

az.plot_rank(many_coefficient_trace, backend_kwargs={'linewidth':0.5, 'markersize':1, 's':1})

The docstring says

These are kwargs specific to the backend being used. For additional documentation check the plotting method of the backend.

, which seems to suggest passing this dictionary is the correct usage. How can this be changed to accept kwargs, or could an example usage be added?

closed time in 2 days

jsevo

PR opened pymc-devs/pymc3

Add Bayesian Additive Regression Trees (BARTs)

This add BARTs to PyMC3 at the API level this looks (almost) like a new distribution (more on this below). Additionally this add a new sampler specifically designed for BART, the PGBART sampler, this is necessary because trees are a very particular kind of discrete distribution (we can also see them as stepwise functions).

The general idea of BARTs is that given a problem of the form y = f(X), we can approximate the unknown function f as a sum of m trees. As trees can easily overfit, BARTs put priors over trees to make each tree only capable of explain a little bit of the data (for example trees tend to be shallow) and thus we must sum many trees to get a reasonable good approximation.

A 1D example.

with pm.Model() as model:
    σ = pm.HalfNormal('σ', 1)
    μ = pm.BART('μ', X, Y, m=50)
    y = pm.Normal('y', μ, σ, observed=Y)

BART_simple_linear_regression_new

The black line is the mean of μ and the band the HDI of μ. As you can see the mean is not a smooth curve because trees are discrete. Notice that this mean is a sum of 50 trees over 2000 posterior draws (2 chains each one of 1000 draws)

This work is the continuation of what @jmloyola did for the GSOC. The main differences are that I reduce most of the trees-code to the essential parts, I try to speed-up things (probably there is still room for improvement) and mainly that I focused on trying to make BART to work inside a probabilistic programming language (I mention this because there is a family of BART methods, in general they are designed with an specific likelihood in mind, and thus they rely on conjugancy). My goal for the BART implementation in PyMC (this will need more PRs) is that BART becomes as flexible as any other distribution, so it can be combined with other distribution to create arbitrary models. At the moment its parameters m and alpha must be floats, not distributions, the main reason is that this is generally the case. There are some reports in the literature saying that putting priors on top of that parameters does not work computationally very well, but this is something I would like to explore.

Some missing features I will like to work on future PRs: Variable selection methods, better test, documentation, store info that could be used for diagnostics. And do some research to better grasp how it behaves for real/complex datasets and way to better select its parameters (loo, CV, priors...)

+718 -5

0 comment

9 changed files

pr created time in 2 days

push eventaloctavodia/pymc3

aloctavodia

commit sha 6a58daaeaf02ae8a67ff74b608db4b1844bf7e92

add tests and minor fixes. Co-authored-by: aloctavodia <aloctavodia@gmail.com> Co-authored-by: jmloyola <jmloyola@outlook.com>

view details

push time in 2 days

push eventaloctavodia/pymc3

aloctavodia

commit sha c3c3929349ec9fb90794994f2ac49c1ab56c7a93

clean docstrings

view details

push time in 2 days

push eventaloctavodia/pymc3

aloctavodia

commit sha 03758a4758f7065adc524b56f57ffc1f20d1b3c1

refactor and clean code

view details

push time in 2 days

push eventaloctavodia/pymc3

aloctavodia

commit sha 51165d4b94d75908f2efcfea3be5bba5379fbd53

improve docstrings

view details

push time in 4 days

push eventaloctavodia/pymc3

aloctavodia

commit sha 3419e70d4da871d540daf6048e94655e12cd3cc4

choose max

view details

push time in 4 days

push eventaloctavodia/pymc3

aloctavodia

commit sha 0ff5833ddfb9d67b81d1892e52aa1ea8bacb4bac

speed-up by fitting a subset of trees per step

view details

push time in 4 days

push eventaloctavodia/pymc3

aloctavodia

commit sha b566d50dc5c702a81fb4d87bab51b3f1f4e1c143

reduce code

view details

push time in 4 days

push eventaloctavodia/Theano-PyMC

Brandon T. Willard

commit sha 9733a2ab8810ca92ac7146ffec0433c8f26312a9

Fix exception handling in as_tensor_variable This change prevents downstream code (e.g. test value computation) from throwing exceptions (e.g. `ValueError`s) that break the logic in `theano.tensor.basic.as_tensor_variable`.

view details

Brandon T. Willard

commit sha 9f56e41882bf43f0b5d7a56ca67ae8b2a350fa7b

Remove redundant AdvancedBoolean*Subtensor classes This change also fixes the `set_subtensor` boolean gradient bug in #105.

view details

Brandon T. Willard

commit sha 53486aae77f16f28060a356f6d0bb5774b9c1a50

Prevent test value computation from complaining about subtensor creation The temporary values created during subtensor broadcast-value computations would cause the test value computation steps to complain/err. This change prevents that.

view details

Brandon T. Willard

commit sha 2352fcec2c67fcab6499c18df222700ef90670c5

Update and refactor logger warnings

view details

Brandon T. Willard

commit sha 74af3ca25eb336428e6ca7b349b363374bd6b729

Rename scratchpad class to Scratchpad

view details

Brandon T. Willard

commit sha 9df70fdad159c1205531202a2dfc14295c0735d6

Move some local optimizations' comments into their docstrings

view details

Brandon T. Willard

commit sha 74ee82b5decee70d31930ea55b40cdc835a3fcdc

Improve variable names and comments/docstring in local_elemwise_fusion_op

view details

Brandon T. Willard

commit sha 21daf4e0fb82b9024ac1a7b57a3a815ba9f1ba68

Validate test values using a tensor's type

view details

Brandon T. Willard

commit sha af703ed29b1483177a81a89b45dd41c31f861cc7

Remove low quality FunctionGraph test relying on stale pickle data This test simply loaded an old example of pickled data. It was both extremely indirect and brittle. Changes to the code base will eventually invalidate nearly all of the old classes used, and whatever the exact issue was, it should've been tested directly.

view details

Brandon T. Willard

commit sha 746b85f42c9ea1d1f9a1a4bb2f6673b9cf588da3

Remove Python 2.x pickle compatibility class and more stale pickle data tests

view details

Brandon T. Willard

commit sha 5194e7e56f9f0bd08d514b19b0f2db3aeb735d51

Remove unused debug-named functions in theano.gof.op

view details

Brandon T. Willard

commit sha 717fd402d0287daeb7875965096dba4d7453ec5a

Refactor test message construction in get_debug_values

view details

Brandon T. Willard

commit sha 747e80fd69495d9ccb301fc2aef7afac17ecb39d

Remove redundant test value check in PureOp._get_test_value

view details

Brandon T. Willard

commit sha ea44b16df4ddab420e1e57846ad3f36a6a813a75

Refactor imports and test design in tests.gof.test_op

view details

Brandon T. Willard

commit sha 9f7a1b69bc78d6185b295540c0f5d315cdbe6cb7

Merge pull request #110 from brandonwillard/enforce-sane-test-values Refactor test value framework so that test value validation is performed up-front.

view details

Brandon T. Willard

commit sha 1d82da43b9013ce2c9fe36f745a0809a7d61a41d

Add a pre-commit config

view details

Brandon T. Willard

commit sha 63fd27b50da285657eb918cb7acdfb016482b471

Update and add missing docstrings in theano.gof.op

view details

Brandon T. Willard

commit sha 03b82c26b1ffddd07d9e1778a703d392ffcebc62

Update and add missing docstrings in theano.gof.graph

view details

Brandon T. Willard

commit sha 0ea1058833e02648ff0fe6bb0f5ad8c8c0122957

Move get_test_value method to the Variable (sub)classes The test value computation steps were also extracted from `PureOp.__call__` and put into a stand-alone `compute_test_value` function.

view details

Brandon T. Willard

commit sha da6ea2fbaaf16c296689524204bac6439bd30a0a

Rename get_debug_values to get_test_values

view details

push time in 5 days

push eventaloctavodia/pymc3

aloctavodia

commit sha 7d54bfa71985494b514c7ebcc9dd3149d45dd33b

clean code add docstring

view details

push time in 5 days

delete branch arviz-devs/arviz

delete branch : violin_plot

delete time in 6 days

push eventarviz-devs/arviz

Osvaldo Martin

commit sha f881be28b1d8e895aacd8c5fc083c9adbc718db3

violinplot remove ticks and spines (#1426) * remove ticks and spines * fix black and add vectorized_to_hex * remove matplotlib color conversion, better fix in a separated PR * update changelog * remove unused import

view details

push time in 6 days

PR merged arviz-devs/arviz

violinplot remove ticks and spines
  • [x] Follows official PR format
  • [x] Includes a sample plot to visually illustrate the changes (only for plot-related functions)
  • [x] Code style correct (follows pylint and black guidelines)
  • [x] Changes are listed in changelog

This was passing silently because with the default arviz styles plots are "despined" (or because nobody use plot_violin, haha)

Old matplotlib old_style

New matplotlib new_style

Old bokeh bokeh_0

New bokeh bokeh_new

Still don't know how to remove the space between subplots in bokeh, @ahartikainen?

+20 -1

1 comment

3 changed files

aloctavodia

pr closed time in 6 days

push eventarviz-devs/arviz

Oriol Abril-Pla

commit sha 99aff55b54cfa9af4d28eb4cc3aad3f215d08186

fix typo in loo_pit (#1418) * fix typo in loo_pit * update changelog * add test * remove changelog duplicated entries * black

view details

Oriol Abril-Pla

commit sha a30e82907487d6b951953856f2c958cd7f698c2b

json dump nested dicts and lists attrs (#1417) * json dump nested dicts and lists * jsonize after loop * update changelog * black

view details

Oriol Abril-Pla

commit sha c9870c8962b6591a715436a078d6b2d917a73ee0

Documentation reorganization (#1406) * inital changes to try pydata sphinx theme * reorganize docs into source folder * restructure api and user guide * start using MyST for md and notebooks * fix gallery generator * add governance and coc links in about page * update azure commands * fix doc warnings * API reference update * black * fix gallery generator and add social links * add home page images in right sidebar * half-working layout it is a little more similar to the current home page but the grid and the title are quite off * get home page body to be sidebar-sidebar width * fix image grid in home page * update changelog

view details

Oriol Abril-Pla

commit sha 12b368f87d2d7d2a1a76644d257b9458ea2580dc

add 404 page to docs (#1423) * add 404 page to docs * center card

view details

Osvaldo Martin

commit sha 0e4e188add3f6a66dba01d8f4cc1a077ccfcf5d4

plot_ppc add observed argument to (un)plot observed data (#1422) * add observed argument to (un)plot observed data * update changelog * fix docstring

view details

Osvaldo Martin

commit sha 6f67d02c29935e3dce866eb33c5d1c75de718f3a

Merge branch 'master' into plot_rank_kwargs

view details

push time in 6 days

delete branch arviz-devs/arviz

delete branch : ppc_obs

delete time in 6 days

push eventarviz-devs/arviz

Osvaldo Martin

commit sha 0e4e188add3f6a66dba01d8f4cc1a077ccfcf5d4

plot_ppc add observed argument to (un)plot observed data (#1422) * add observed argument to (un)plot observed data * update changelog * fix docstring

view details

push time in 6 days

PR merged arviz-devs/arviz

plot_ppc add observed argument to (un)plot observed data
  • [x] Follows official PR format
  • [x] Includes new or updated tests to cover the new feature
  • [x] Code style correct (follows pylint and black guidelines)
  • [x] Changes are listed in changelog

I also made a few other changes in bokeh, specially to use the matplotlib color and to make the scatter plot a little fit faster.

+141 -107

1 comment

6 changed files

aloctavodia

pr closed time in 6 days

issue openedbambinos/bambi

Remove Bayesian label from examples

All the examples names are Bayesian "something" which is redundant in a Bayesian library :-)

created time in 6 days

push eventarviz-devs/arviz

aloctavodia

commit sha a4eab9117e18fe0e6e9c5034f6383cb8c761e70e

remove unused import

view details

push time in 6 days

push eventarviz-devs/arviz

aloctavodia

commit sha 331ecb0c9760931e0cd1bacca66807fe327fd075

update changelog

view details

push time in 6 days

push eventarviz-devs/arviz

aloctavodia

commit sha c40e962d0dc41e568cd86b9822eae1ef1b4dcdd5

remove matplotlib color conversion, better fix in a separated PR

view details

push time in 6 days

push eventarviz-devs/arviz

aloctavodia

commit sha 2ea6b870da36b3dda5edbc47b0f4cfda6c939b90

fix docstring

view details

push time in 6 days

delete branch aloctavodia/bambi

delete branch : plot_prior_args

delete time in 6 days

push eventbambinos/bambi

Osvaldo Martin

commit sha 3d0eebbbea4ddbaa8ce414a30049c9fcfe955e79

add ploting arguments (#255)

view details

push time in 6 days

push eventarviz-devs/arviz

aloctavodia

commit sha e85e864b08565210102a5cf2ab6db786916df3bf

fix black and add vectorized_to_hex

view details

push time in 6 days

PR opened arviz-devs/arviz

violinplot remove ticks and spines
  • [x] Follows official PR format
  • [x] Includes a sample plot to visually illustrate the changes (only for plot-related functions)
  • [x] Code style correct (follows pylint and black guidelines)
  • [ ] Changes are listed in changelog

This was passing silently because with the default arviz styles plots are "despined" (or because nobody use plot_violin, haha)

Old matplotlib old_style

New matplotlib new_style

Old bokeh bokeh_0

New bokeh bokeh_new

Still don't know how to remove the space between subplots in bokeh, @ahartikainen?

+19 -1

0 comment

2 changed files

pr created time in 6 days

create barncharviz-devs/arviz

branch : violin_plot

created branch time in 6 days

push eventarviz-devs/arviz

aloctavodia

commit sha c94c7ef922b97aeb9fe0d4f2491d40ac5ba4f84c

update changelog

view details

push time in 6 days

push eventaloctavodia/pymc3

aloctavodia

commit sha ac96b1a38004884090586c6e4aacc0cff11ed821

use Rockova prior, refactor prior leaf prob computaion

view details

push time in 8 days

push eventaloctavodia/pymc3

aloctavodia

commit sha 6700a74ca4a6567a58265a4e1b9d9a3f56487c8e

use a low number of max_stages for the first iteration, remove not necessary errors

view details

push time in 8 days

PR opened bambinos/bambi

plot_priors: add ploting arguments
+71 -4

0 comment

1 changed file

pr created time in 9 days

create barnchaloctavodia/bambi

branch : plot_prior_args

created branch time in 9 days

push eventaloctavodia/bambi

Osvaldo Martin

commit sha 3b53b1a0254035c847a307f78fd582ee6d24070d

update black (#254) * update black * update create_testenv

view details

push time in 9 days

PR opened arviz-devs/arviz

plot_ppc add observed argument to (un)plot observed data
  • [x] Follows official PR format
  • [ ] Includes a sample plot to visually illustrate the changes (only for plot-related functions)
  • [ ] New features are properly documented (with an example if appropriate)?
  • [ ] Includes new or updated tests to cover the new feature
  • [ ] Code style correct (follows pylint and black guidelines)
  • [ ] Changes are listed in changelog
+140 -107

0 comment

5 changed files

pr created time in 9 days

create barncharviz-devs/arviz

branch : ppc_obs

created branch time in 9 days

PR opened arviz-devs/arviz

[WIP] geweke inspired diagnostic and plot
  • [ ] Follows official PR format
  • [ ] Includes a sample plot to visually illustrate the changes (only for plot-related functions)
  • [ ] New features are properly documented (with an example if appropriate)?
  • [ ] Includes new or updated tests to cover the new feature
  • [ ] Code style correct (follows pylint and black guidelines)
  • [ ] Changes are listed in changelog

This is mostly for discussion, still unsure if this is something we want to add to ArviZ, or what exactly this is adding to the current set of diagnostic tools. I was working on updating our geweke diagnostic but instead I took a detour. This proposed diagnostics (called geweke_like in the absence of a better name) concatenates all chains and split them in equal size portions. Then it compares them pairwise (all against all) by computing the difference of the mean divided by their pooled variances. This is essentially a Welch's t statistics. The computed z_scores are expected to be distributed as a standard normal distribution. Notice this is not trying to asses the need for burn-in.

A few examples using the centered_eight and non_centered_eight idata

plot_geweke_like(centered, kind="forest")

gforest_centered_

plot_geweke_like(non_centered, kind="forest")

gforest_non_centered_

plot_geweke_like(centered)

gscatter_centered

plot_geweke_like(non_centered)

gscatter_non_centered

+172 -0

0 comment

1 changed file

pr created time in 9 days

create barncharviz-devs/arviz

branch : geweke_like

created branch time in 9 days

push eventarviz-devs/arviz

aloctavodia

commit sha 8b8299d6c98a4ab489d3ce6ef8d8a02820cf38c8

add example

view details

aloctavodia

commit sha 68193c5ca43ad49792b7886891edd42b727d67a3

update changelog

view details

push time in 9 days

delete branch aloctavodia/bambi

delete branch : black_v

delete time in 9 days

push eventbambinos/bambi

Osvaldo Martin

commit sha 3b53b1a0254035c847a307f78fd582ee6d24070d

update black (#254) * update black * update create_testenv

view details

push time in 9 days

PR merged bambinos/bambi

update black
+9 -2

0 comment

3 changed files

aloctavodia

pr closed time in 9 days

delete branch arviz-devs/arviz

delete branch : doc_theme

delete time in 9 days

push eventarviz-devs/arviz

Oriol Abril-Pla

commit sha c9870c8962b6591a715436a078d6b2d917a73ee0

Documentation reorganization (#1406) * inital changes to try pydata sphinx theme * reorganize docs into source folder * restructure api and user guide * start using MyST for md and notebooks * fix gallery generator * add governance and coc links in about page * update azure commands * fix doc warnings * API reference update * black * fix gallery generator and add social links * add home page images in right sidebar * half-working layout it is a little more similar to the current home page but the grid and the title are quite off * get home page body to be sidebar-sidebar width * fix image grid in home page * update changelog

view details

push time in 9 days

PR merged arviz-devs/arviz

Documentation reorganization

Description

Related to #1331. This PR intends to reorganize the documentation in order to make it easier to extend it. There are several important changes to do this:

  • Use MyST for notebooks and markdown files.
    • MyST (and MyST-NB) is a sphinx parser for markdown that also extends markdown in order to allow using sphinx directives from markdown. This allows to keep the schema written in markdown and to use sphinx directive from the example jupyter notebooks. Now for example, notebooks can be referred to using ref links and they can use ref, class, func... links to refer to other pages.
  • Change the theme to pydata_sphinx_theme to get automatic tables of contents for both page and section at the same time (one in each sidebar)
  • Reorganize the folders in which the documentation is stored.

todo list:

  • ~Fix gallery generator~
  • ~Think about pulling relevant docs (i.e. CONTRIBUTING is already included in the docs) such as governance, code of conduct, readme?...~ I think it's fine for now

EDIT: preview: https://oriolabril.github.io/arviz/index.html

Checklist

<!-- Feel free to remove check-list items aren't relevant to your change -->

  • [x] Follows official PR format
  • [x] New features are properly documented (with an example if appropriate)?
  • [x] Code style correct (follows pylint and black guidelines)
  • [x] Changes are listed in changelog

<!-- Also, please consider reading the contributing guidelines and code of conduct carefully before submitting the PR. They are available at

  • https://github.com/arviz-devs/arviz/blob/master/CONTRIBUTING.md
  • https://github.com/arviz-devs/arviz/blob/master/CODE_OF_CONDUCT.md

We understand that PRs can sometimes be overwhelming, especially as the reviews start coming in. Please let us know if the reviews are unclear or the recommended next step seems overly demanding, if you would like help in addressing a reviewer's comments, or if you have been waiting too long to hear back on your PR. -->

+656 -446

7 comments

71 changed files

OriolAbril

pr closed time in 9 days

push eventaloctavodia/bambi

aloctavodia

commit sha 543b4464c0e710f37acd7fc8919e623c60fcd0a5

update create_testenv

view details

push time in 9 days

PR opened bambinos/bambi

update black
+8 -2

0 comment

2 changed files

pr created time in 9 days

create barnchaloctavodia/bambi

branch : black_v

created branch time in 9 days

push eventaloctavodia/bambi

Osvaldo Martin

commit sha 6404d88f97d359d221c32f00c40d79b2b3f94ba9

add bambi info to idata attributes (#253)

view details

push time in 9 days

push eventbambinos/bambi

Osvaldo Martin

commit sha 6404d88f97d359d221c32f00c40d79b2b3f94ba9

add bambi info to idata attributes (#253)

view details

push time in 9 days

PR merged bambinos/bambi

add bambi info to idata attributes

This add the "modeling_interface" and "modeling_interface_version" attributes to the InferenceData objects returned by model.fit, model.prior_predictive and model.posterior_predictive. To avoid possible confusion and help reproducibility the attributes "inference_library" and "inference_library_version" are those of the backend (i.e. pymc3)

+18 -3

0 comment

2 changed files

aloctavodia

pr closed time in 9 days

push eventarviz-devs/arviz

aloctavodia

commit sha 19f6f1722521c1609e39c4d4e8db926698ca2d9f

black

view details

push time in 10 days

PR opened arviz-devs/arviz

plot_rank: add ref_line, bar, vlines and marker_vlines kwargs

Fix #1396

  • [x] Follows official PR format
  • [ ] Includes a sample plot to visually illustrate the changes (only for plot-related functions)
  • [ ] New features are properly documented (with an example if appropriate)?
  • [ ] Includes new or updated tests to cover the new feature
  • [ ] Code style correct (follows pylint and black guidelines)
  • [ ] Changes are listed in changelog
+117 -82

0 comment

5 changed files

pr created time in 10 days

create barncharviz-devs/arviz

branch : plot_rank_kwargs

created branch time in 10 days

delete branch arviz-devs/arviz

delete branch : kde_bw

delete time in 10 days

push eventarviz-devs/arviz

Osvaldo Martin

commit sha 86b4aa9e98c98b115aeb09df43ef1c7c9237a685

add warning when bw estimation fails and add failsafe value (#1415) * add warning when bw estimation fails and add failsafe value * only check for finite values

view details

push time in 10 days

PR merged arviz-devs/arviz

add warning when bw estimation fails and add failsafe value

This should make plot_kde to works even when the passed array has a single repeated value, and it will report a warning.

+8 -3

1 comment

1 changed file

aloctavodia

pr closed time in 10 days

PullRequestReviewEvent

push eventaloctavodia/Theano-PyMC

Brandon T. Willard

commit sha ba126742c32f82b7e20f87f6b9f1e4fa42d7f3cd

Convert boolean indices to integers when determining broadcast pattern

view details

push time in 10 days

PR opened bambinos/bambi

add bambi info to idata attributes

This add the "modeling_interface" and "modeling_interface_version" attributes to the InferenceData objects returned by model.fit, model.prior_predictive and model.posterior_predictive. To avoid possible confusion and help reproducibility the attributes "inference_library" and "inference_library_version" are those of the backend (i.e. pymc3)

+18 -3

0 comment

2 changed files

pr created time in 11 days

create barnchaloctavodia/bambi

branch : attributes

created branch time in 11 days

push eventaloctavodia/bambi

Osvaldo Martin

commit sha f613b838c0505fe1cd162a0eff8ac72e4f514b48

add posterior predictive method (#250) * add posterior predictive sampling * update changelog and apply black * fix lint

view details

push time in 11 days

delete branch aloctavodia/bambi

delete branch : posterior_predictive

delete time in 11 days

push eventbambinos/bambi

Osvaldo Martin

commit sha f613b838c0505fe1cd162a0eff8ac72e4f514b48

add posterior predictive method (#250) * add posterior predictive sampling * update changelog and apply black * fix lint

view details

push time in 11 days

PR merged bambinos/bambi

add posterior predictive method

Fix #212

+71 -0

0 comment

3 changed files

aloctavodia

pr closed time in 11 days

push eventaloctavodia/bambi

aloctavodia

commit sha 2ae9d200bf7a4f0c7b65c1a72a47c4e16b8fa0c7

fix lint

view details

push time in 11 days

push eventaloctavodia/bambi

aloctavodia

commit sha a40e546a937fdf464f8da44018262e447dbaf262

update changelog and apply black

view details

push time in 11 days

PR opened bambinos/bambi

add posterior predictive method

Fix #212

+70 -0

0 comment

2 changed files

pr created time in 11 days

create barnchaloctavodia/bambi

branch : posterior_predictive

created branch time in 11 days

push eventaloctavodia/bambi

Osvaldo Martin

commit sha e78f6105b0cc77f46e68aa3b3232a04f1a688aa1

Replace samples argument with draws, add deprecation warning (#247) * replace samples argument with draws, add deprecation warning * fix typo * fix tests * black * update changelog

view details

push time in 11 days

issue closedbambinos/bambi

Changelog is outdated

closed time in 11 days

aloctavodia

delete branch aloctavodia/bambi

delete branch : draws

delete time in 11 days

push eventbambinos/bambi

Osvaldo Martin

commit sha e78f6105b0cc77f46e68aa3b3232a04f1a688aa1

Replace samples argument with draws, add deprecation warning (#247) * replace samples argument with draws, add deprecation warning * fix typo * fix tests * black * update changelog

view details

push time in 11 days

PR merged bambinos/bambi

Replace samples argument with draws, add deprecation warning

closes #237

+72 -67

1 comment

11 changed files

aloctavodia

pr closed time in 11 days

issue closedbambinos/bambi

Replace the argument “samples” by “draws”.

This will be consistent with the use in PyMC and ArviZ.

optionally we could add somewhere a short discussing about how draws and samples are different, but not sure if it is necessary to be part of the Bambi documentation.

closed time in 11 days

aloctavodia

push eventaloctavodia/pymc3

Maxim Kochurov

commit sha 5e305543a13d248cb8cd67522527c586bde45165

Circular kernel (#4082) * Adding Cirdular kernel * adding an example notebook * ad docs, append to release notes * update circular example * Update RELEASE-NOTES.md Co-authored-by: Alexandre ANDORRA <andorra.alexandre@gmail.com> * review nb * fix typos * circular update only * update kernels and covs nb * fix typos * bound->period for consistency with Periodic * update plots, notebooks * fix typo in doc * fix even more typos * follow pre-commit advice * typo fix * update notebooks * nitpics * fix typo Co-authored-by: Alexandre ANDORRA <andorra.alexandre@gmail.com>

view details

Marco Gorelli

commit sha 4fff38bf7d503f10f6316456aadaa572d0e45d03

nbqa black 5 notebooks (#4165)

view details

push time in 11 days

push eventaloctavodia/Theano-PyMC

Brandon T. Willard

commit sha 6c8689ccd4d52a7797f4955556c7c2c924145330

Remove unnecessary self argument in super call

view details

Brandon T. Willard

commit sha d05ab606d37ec035da441fb385b8959e322e55bf

Add a test for jaxification of a shared variable

view details

Dan F-M

commit sha 5de03387985b7eea5bc155d7c2250beb17a5dd7a

adding GitHub actions workflow

view details

Brandon T. Willard

commit sha cb2e0340f59f3a0e809e4ef95f06423d60ddfc98

Remove unnecessary dictionary from CumOp.perform

view details

Brandon T. Willard

commit sha 6bf0d06d0b40a8e2d251e5578c36c5c4f1c85610

Add TensorSharedVariable support to get_vector_length

view details

Brandon T. Willard

commit sha 44aa0fb5c2af86dc6f02fc44c5160edfd56897a8

Remove unnecessary ndim property and use order property in UnravelIndex

view details

Brandon T. Willard

commit sha 8a7f59e1011e92d25e75c1287ff83c38e51d79d2

Add an option to remove all optimizations and ignore result dtypes in JAX testing

view details

Brandon T. Willard

commit sha 9ea670cad7c5a9e6cacb7a35b6f45152483ae913

Remove type check on JAX multi-output tests

view details

Brandon T. Willard

commit sha d5b9940dccb15ba06e80f176d97e253b0f2111dd

Use random input instead of NumPy empty values in JAX test `np.empty` was producing values at the edge of numerical zero. These values would apparently cause `jax` to truncate something during and intermediate step of a comparison operation like `array > 0`, which would produce a result different from standard NumPy's.

view details

Brandon T. Willard

commit sha e464ba491fbf2a7f98fb899d3dd06e78ba842871

Make sure ifelse is imported before the JAX test is run

view details

Brandon T. Willard

commit sha fe0365add873b3dc87c9d67d21cb6de90cb9e4f0

Implement new JAX conversions for theano.tensor.extra_ops

view details

Brandon T. Willard

commit sha 8924d29436b729647ed1037e26270907b555d958

Merge pull request #80 from brandonwillard/add-cumop-jaxification Add JAX conversions for theano.tensor.extra_ops

view details

Dan Foreman-Mackey

commit sha 64ae1f76655e9784a640ea6807bbc70eff39f2e2

Fix coveralls for pull requests (#86)

view details

Brandon T. Willard

commit sha 295ea57d0f8bc585a5d1f4109815e2bce47d8a4b

Prevent CI from running Python tests for non-Python changes

view details

Michael Osthege

commit sha 0421b6f5b49c84664d8bc0293820b1628b2f5e66

Remove import of test suite from main __init__ - The 6-year old comment indicated that the import was done because of some GPU configuration, without providing any details - The import actually imported the first thing called "tests" that is found in the PATH

view details

Brandon T. Willard

commit sha 5551061f013dda7ef6a8971959c016deb9358483

Fix broken gpuarray imports from tests.tensor

view details

Brandon T. Willard

commit sha f683d1ac735355f01bdb91e0b3a768360ead8cd7

Remove unused truedot.py module with unnecessary tests imports

view details

Brandon T. Willard

commit sha f8ce40ae7ecef2a061912062519a0ccb78be7ad0

Relocate str_diagnostic function from tests to theano.compile.debugmode

view details

Brandon T. Willard

commit sha 2957ed4e005171f30357e85da5b8cefe7084e811

Move PdbBreakpoint Op from tests to the theano package

view details

Brandon T. Willard

commit sha 21a993ea690b3c5123ae4d692c1b484180b3df79

Remove theano.sandbox.cuda containing a hidden pytest import

view details

push time in 11 days

push eventaloctavodia/bambi

aloctavodia

commit sha 59d066265d098e5a7786cc3cab6eae7828409a88

update changelog

view details

push time in 12 days

push eventaloctavodia/bambi

Osvaldo Martin

commit sha 2a3559f6ce6c2f062a85db1435926445b7ac5057

update changelog (#248)

view details

Osvaldo Martin

commit sha 6ad8d8703efa5aba2ca323d99a294636d1c7d222

merge union (#249)

view details

aloctavodia

commit sha 543045d8b2c300296b5f68630705483ea5d58a17

replace samples argument with draws, add deprecation warning

view details

aloctavodia

commit sha 28abdcbc59261c6f6a3843929cfe54f2f07c43ab

fix typo

view details

aloctavodia

commit sha 0d3427f69e804b6fdbff256c6c727a398593a208

fix tests

view details

aloctavodia

commit sha 60a7c39eef880b9f6a9a7980e0d50273ca5832e1

black

view details

push time in 12 days

push eventaloctavodia/bambi

Osvaldo Martin

commit sha 6ad8d8703efa5aba2ca323d99a294636d1c7d222

merge union (#249)

view details

push time in 12 days

delete branch aloctavodia/bambi

delete branch : changelog2

delete time in 12 days

push eventbambinos/bambi

Osvaldo Martin

commit sha 6ad8d8703efa5aba2ca323d99a294636d1c7d222

merge union (#249)

view details

push time in 12 days

PR merged bambinos/bambi

add gitattributes
+1 -0

0 comment

1 changed file

aloctavodia

pr closed time in 12 days

PR opened bambinos/bambi

merge union
+1 -0

0 comment

1 changed file

pr created time in 12 days

more