Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Tests for plotting functions and methods, other increases in test coverage #384

Merged
merged 36 commits into from Mar 30, 2020

Conversation

ulupo
Copy link
Collaborator

@ulupo ulupo commented Mar 30, 2020

Reference issues/PRs
Addresses several remaining gaps in coverage (but still not the ones indicated in #182).

Types of changes

  • Bug fix (non-breaking change which fixes an issue)
  • New feature (non-breaking change which adds functionality)
  • Breaking change (fix or feature that would cause existing functionality to change)

Description

  • Adds tests for all PlotterMixins or other classes containing plot methods. These are just dummy tests implementing trivial calls to fit_transform_plot or plot.
  • Updates .coveragerc file so that gtda.images is included in coverage reports, gtda.utils._docs and gtda.utils.intervals are excluded, and all test directories are excluded.
  • Adds simple tests for Filtering.
  • Adds simple shape tests for BettiCurve and PersistenceLandscape.
  • Minor typo fixes and other code style improvements.

Checklist

  • I have read the guidelines for contributing.
  • My code follows the code style of this project. I used flake8 to check my Python changes.
  • My change requires a change to the documentation.
  • I have updated the documentation accordingly.
  • I have added tests to cover my changes.
  • All new and existing tests passed. I used pytest to check this on Python tests.

Umberto added 30 commits March 23, 2020 12:46
- Rename test_features -> test_features_representations
- Create parametrized dummy tests for fit_transform_plot method of transformers in representations.py
- Improve structure and range of tests in test_preprocessing in a similar way
- Create dummy tests of shape for BettiCurve and PersistenceLandscape
- Linting and code style changes elsewhere
@ulupo ulupo requested a review from wreise March 30, 2020 07:35
@ulupo
Copy link
Collaborator Author

ulupo commented Mar 30, 2020

@wreise @gtauzin one downside of having plot methods which not only create figures but also show them is that locally running tests causes one tab per plot to be automatically opened by my browser. Have you encountered/resolved this?

@ulupo
Copy link
Collaborator Author

ulupo commented Mar 30, 2020

The failures in the manylinux builds seem to be related to a problem with the boost website. I wouldn't worry too much about it yet, hopefully it will be up and running soon.

gtda/diagrams/representations.py Show resolved Hide resolved


@pytest.mark.parametrize('hom_dim_ix', [0, 1])
def test_fit_transform_plot_1(hom_dim_ix):
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Would you mind giving a more descriptive name, or a comment for future selves?
If i understand correctly, we plot a single homology dimension.

Copy link
Collaborator Author

@ulupo ulupo Mar 30, 2020

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Some classes have the homology_dimension_ix parameter, others homology_dimensions. I could not find a short convincing name, but test_fit_transform_plot_one_hom_dim for this one, and test_fit_transform_plot_single_many_hom_dims for the next, could work with me.

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

👍

gtda/diagrams/tests/test_features_representations.py Outdated Show resolved Hide resolved
f'All homology dimensions must be in '
f'self.homology_dimensions_ which is '
f'{self.homology_dimensions_}. {dim} is not.')
f"All homology dimensions must be in "
Copy link
Collaborator Author

@ulupo ulupo Mar 30, 2020

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm starting to use these quotation marks consistently in sentence-like strings as they are safe when apostrophes have to be used inside as well.

- Rename some test functions
- Implement a solution to the automatic tab opening problem, based on setting the default value of plotly.io.renderers in test files
@ulupo
Copy link
Collaborator Author

ulupo commented Mar 30, 2020

@gtauzin this is ready to merge and the new coverage percentage is between 93% and 94%. Due to issues at the boost website's end (Forbidden errors when following the tar.gz download link), we are delaying merging until the latter works as expected again.

@ulupo ulupo merged commit c78de1e into giotto-ai:master Mar 30, 2020
@ulupo ulupo deleted the simple_plotting_tests branch March 30, 2020 12:26
ulupo added a commit that referenced this pull request Apr 8, 2020
* Add twine check to CI, attempt fix in README (#379)

* Tests for plotting functions and methods, other increases in test coverage (#384)

* Vectorization2 (#378)

* Fix azure on windows (#388)

* Allow more general input to ripser when metric='precomputed', improve/refactor check_point_clouds and add tests (#386)

* Add citing entry to README.rst and doc/faq.rst (#389)

* Fix markdown subsection formatting (#390)

* Prepare release 0.2.1 (#391)

Co-authored-by: Anibal M. Medina-Mardones <ammedmar@gmail.com>
Co-authored-by: REDS institute <reds-heig@users.noreply.github.com>
Co-authored-by: Guillaume Tauzin <guillaumetauzin.ut@gmail.com>
wreise added a commit that referenced this pull request Apr 8, 2020
* Add temporary patch to azure-pipelines.yml (#369)

Get around Azure DevOps issues by following a suggestion by Sean Law: https://twitter.com/seanmylaw/status/1239332444865773568

* Remove metatransformers (#368)

Signed-off-by: Guillaume Tauzin <guillaumetauzin.ut@gmail.com>

* Extend plotting functionalities to more transformers and other modules (#362)

* start plotting for images

* Add the image plotting  file

* Add an imagePlotMixin subclassing the ImagePlot for theimage submodule, add persistenceImage plots

* Add plot_persistence_image to the package

* Add plotting functionalities to filtrations in the image module and forbetticurves

* Preprocessing covered by plots

* Add basic plotting functions for different vectorisation  methods

* Incorporate plotting for diagram representations and images/preprocessing

* Change the docstring for ImagePlotterMixin and add fail on 3d  images

* Correct the samplingsin features, and titles

* Small wording fix

* Change dosctrings for plot in diagrams/features

* Fix typo in variable name

* Change plotting of  heatkernel to the same as persistenceimage, use  samplings

* Cast  images to floats, so as they can be always plotted

* Add a Homology and PointCloud Mixin

* Use the new PlotMixins

* Add todo in Forgetdimensions, in relation to plotting

* Add plotting to point_cloudsrescaling, correctimports

* Wording changes

* Correct n_values to n_bins and  add 's' to n_homologydimensionS

* Correct spelling mistakes

* Change the name of ...PlotterMixin to ...Plotter

* Make SlidingWindowEmbedding a PointCloudPlotter

* Fix typoin import

* Fix linting

* Correct the  way the distance matrices are displayed, and add the  SlidingWindow.plot method

* Change behaviour of transform_plot to return the 0-th sample

* Docstring fixes

* Create diagrams/representations.py, and migrate amplitude to features.py

This change reflects the new documentation structure and the difference now drawn there between feature creation for diagrams (scalars) and vector representations/vectorizations. These changes are mostly invisible to the user.

* Remove unused identity function

* Rename HomologyPlotter as DiagramPlotter

* Add homology_dimensions kwarg to DiagramPlotter, revise docstring

* Small docstring fixes

* Try adding triggers in the yaml

* Remove excessive blank line

* Revert "Change behaviour of transform_plot to return the 0-th sample"

This reverts commit 2fa0209.

* Fix docstring for PlotterMixin

* Further docstring refinements in `PlotterMixin`

* Remove `DiagramPlotter`, differentiate between implementations of `plot`

Also make plot methods static in classes in `gtda.homology` and in ForgetDimensino

* Correctly import PlotterMixin

* Correct the plotting in the lorentz notebook

* Remove PointCloudPlotter and refine plot methods for point clouds

* Thorough refactoring of plotting API

- Eliminate some top-level functions for diagram representations, leave only betti_curves and betti_surfaces
- Remove ImagePlotter mixin and plotting/_plot.py, create top-level plot_heatmap function
- Rename plotting/diagrams.py to plotting/diagrams_representations.py
- Rename plotting/image.py to plotting/images.py
- Rename plotting/homology.py to plotting/persistence_digrams.py
- Fix problem with Binarizer coming from now-removed normalize parameter
- Refactor notebooks to use plotting API
- MIscellaneous small changes elsewhere

The philosophy behind the file changes in gtda/plotting is that file names now reflect data structures and do not try to follow the names of subpackages like diagrams, image, etc.

* Grammar

* Fix default parameter for homology_dimension_ix in HeatKernel

* Change plot_diagrams to allow ForgetDimension to work

* Allow _subdiagrams to better handle infs of different kind

* Add inline comment for future selves

* Write check_diagrams docstring and make it convert to float, revert previous changes to _subdiagrams

* Pass copy=False to astype in check_diagrams

Co-authored-by: Guillaume Tauzin <guillaume.tauzin@epfl.ch>
Co-authored-by: Guillaume Tauzin <guillaumetauzin.ut@gmail.com>
Co-authored-by: Umberto Lupo <46537483+ulupo@users.noreply.github.com>
Co-authored-by: Umberto <u.lupo@l2f.ch>

* Glossary entry: distance, inner products & kernels (#370)

* Update P landscapes

* Add distances, inner products and kernels glossary entry

* Update after  W's comments

Signed-off-by: ammedmar <ammedmar@gmail.com>
Co-authored-by: ammedmar <anibal@medina-mardones.com>

* Add reference for images submodules (#372)

* Add reference for images submodules

Signed-off-by: Guillaume Tauzin <guillaumetauzin.ut@gmail.com>

* Move hera to submodule (#371)

* delete hera repository

* Add hera as submodule

* Remove unnecessary branch for submodule

* Add parameter max_bids_per_round when computing wasserstein

* Fix an inconsistency with wasserstein implementation

In Wasserstein, by default epsilon_common_ratio is equal 5, and it can be
equal 0, which our bindings did not allow

* Add test for python bindings of wasserstein

Add method to retrieve what wasserstein consider to be infinity as an input

* Add test for python bindings of bottleneck

* Add fixes done by author after I contact him

Signed-off-by: julian <julian.burellaperez@heig-vd.ch>

* Miscellaneous refinements to GitHub pages (#363)

* Remove code fetching logo as it is now stored in doc/images

* Fix logo in examples directory

* Partial rewriting of contributing page

* Remove instructions for building documentation

* Simplify path in Makefile

* Remove documentation job in azure-pipelines.yml

* Add @wreise to authors in doc/conf.py

* Remove paper citation from FAQ

* Rename Glossary to Theory Glossary

* Linting in images subpackage

* Add PlotterMixin to base.rst

* Update FAQ

* Improve README with pointers to GH pages

* Implement @wreise's review suggestions

* Fix links to svgs

* Remove remaining mentions of plotting.py

* Rename basic to tutorials and advanced to examples

* Fixes in jupyter notebooks in examples folder

* Add plotting in docs

* Fixes in glossary

* Correct the VBox display

* Add TODO instead of empty reference in Cech docstring

* Fix references in docstrings

* Reintroduce scrollZoom in mapper_quickstart

* Improve landing page and move Lewis' blog post to notebooks/tutorials.rst

Co-authored-by: Wojciech Reise <reisewojtus@gmail.com>

* Pcs different sizes (#318)

* Add a check_point_clouds fct in utils/validation, and apply it throughout classes in ``homology/simplicial.py``

* Adapt and add tests

* Make the output of ImageToPointCloud variable in size,  and adjust the tests (modify the expected values)

* Add mapper and utils to global __init__

* Revise use of check_array throughout

* Linting and code clarity improvements throughout

* Fix some incorrect links to GH pages

* Revert to trivial slicing in transform_plot method of PlotterMixin, to cover case of list input

Co-authored-by: Umberto <u.lupo@l2f.ch>

* Release of v0.2.0 (#375)

* Bump version to 0.2.0, add release notes

* Improve LaTeX rendering in mapper_quickstart.ipynb

* Add @rth as a contributor and code author for v0.2.0 (#376)

* Add @rth to code authors and list of contributors for v0.2.0

* [WIP] Add twine check to CI, attempt fix in README (#379)

* Add twine check to azure-pipelines.yml

* Fix formatting and duplicate issues in README.rst

* Tests for plotting functions and methods, other increases in test coverage (#384)

* Remove images subpackage from code omitted from coverage

* Add tests for Filtering class

* Add tests in diagrams subpackage, linting

- Rename test_features -> test_features_representations
- Create parametrized dummy tests for fit_transform_plot method of transformers in representations.py
- Improve structure and range of tests in test_preprocessing in a similar way
- Create dummy tests of shape for BettiCurve and PersistenceLandscape
- Linting and code style changes elsewhere

* Add fit_transform_plot test for GraphGeodesicDistance

* Add fit_transform_plot tests in gtda.homology

* Fix typo in voids_on_the_plane notebook

* Add fit_transform_plot tests for images subpackage

* Add fit_transform_plot tests for point_clouds subpackage

* Add test for SlidingWindow.plot

* Linting

* Exclude gtda.utils._doc and gtda.utils.intervals from coveragerc

* Implement @wreise's suggestions

- Rename some test functions
- Implement a solution to the automatic tab opening problem, based on setting the default value of plotly.io.renderers in test files

* Upload coverage report only from Azure macOS job

* Exclude test folders from coverage reports

* Vectorization2 (#378)

* Update P landscapes

* Add distances, inner products and kernels glossary entry

Signed-off-by: ammedmar <anibal@medina-mardones.com>

* Fix azure on windows (#388)

* Update ccache

* Update variable used to indicate boost headers location

Official answer from boost devs: https://discourse.cmake.org/t/findboost-difference-between-boost-include-dir-and-boost-include-dirs/855

* Add install boost step in Azure CI, which downloads boost

Signed-off-by: julian <julian.burellaperez@heig-vd.ch>

* Allow more general input to ripser when metric='precomputed', improve/refactor check_point_clouds and add tests (#386)

* Remove boolean mask on infinite values in VietorisRipsPersistence

* Refactoring of check_point_clouds

- Rename distance_matrix kwarg to distance_matrices
- Add checks for square input when distance_matrices is set to True
- More comprehensive/friendly error messages and warnings

* Increase generality of ripser function when metric == 'precomputed'

Sklearn's pairwise_distances is no longer called when metric='precomputed'.

This is because pairwise_distances checks for positivity and finite input in that case, but this is unnecessarily restrictive for ripser. 

pairwise_distances also checks for square input so this is no longer checked in the case metric='precomputed'.  But the ripser function is not meant for use in isolation. It is meant to be called by a VietorisRipsPersistence instance, and input validation can be offloaded to that class. In particular, 9bab872 introduces checks for square input.

* Add tests for check_point_clouds

* Refresh docstrings for VietorisRipsPersistence

- Mention "adjacency matrices of weighted undirected graphs" as well as distance matrices
- Mention compatibility with filtration in docstrings for fit and transform

* Add citing entry to README.rst and doc/faq.rst (#389)

* Add citation section in README

* Add citation info to FAQ page in docs

Signed-off-by: Guillaume Tauzin <guillaumetauzin.ut@gmail.com>

Co-authored-by: Umberto Lupo <46537483+ulupo@users.noreply.github.com>

* Fix markdown subsection formatting (#390)

* Fix citing giotto-tda section
* Fix Important link section

* Prepare release 0.2.1 (#391)

* Make the docs and push to a test branch in gtda-docs

* Fix indentation

Co-authored-by: Umberto Lupo <46537483+ulupo@users.noreply.github.com>
Co-authored-by: Guillaume Tauzin <guillaumetauzin.ut@gmail.com>
Co-authored-by: Guillaume Tauzin <guillaume.tauzin@epfl.ch>
Co-authored-by: Umberto <u.lupo@l2f.ch>
Co-authored-by: Anibal M. Medina-Mardones <ammedmar@gmail.com>
Co-authored-by: ammedmar <anibal@medina-mardones.com>
Co-authored-by: REDS institute <reds-heig@users.noreply.github.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

2 participants