diff --git a/doc/library.rst b/doc/library.rst index 19787b844..2737bd8ae 100644 --- a/doc/library.rst +++ b/doc/library.rst @@ -114,5 +114,5 @@ What's new .. include:: release.rst - :start-after: Release 0.2.2 - :end-before: Release 0.2.1 + :start-after: Release 0.3.0 + :end-before: Release 0.2.2 diff --git a/doc/modules/homology.rst b/doc/modules/homology.rst index f6ba5db68..e0795b02f 100644 --- a/doc/modules/homology.rst +++ b/doc/modules/homology.rst @@ -20,6 +20,12 @@ Undirected simplicial homology Directed simplicial homology ---------------------------- +.. currentmodule:: gtda + +.. autosummary:: + :toctree: generated/homology/ + :template: class.rst + homology.FlagserPersistence Cubical homology diff --git a/doc/release.rst b/doc/release.rst index a27b903a3..cc8f8b00b 100644 --- a/doc/release.rst +++ b/doc/release.rst @@ -4,6 +4,187 @@ Release Notes .. _stable: +************* +Release 0.3.0 +************* + +Major Features and Improvements +=============================== + +This is a major release which adds substantial new functionality and introduces several improvements. + +Persistent homology of directed flag complexes via ``pyflagser`` +---------------------------------------------------------------- + +- The ``pyflagser`` package (`source `_, `docs `_) is now an official dependency of ``giotto-tda``. +- The ``FlagserPersistence`` transformer has been added to ``gtda.homology`` (`#339 `_). It wraps ``pyflagser.flagser_weighted`` to allow for computations of persistence diagrams from directed or undirected weighted graphs. A `new notebook `_ demonstrates its use. + +Edge collapsing and performance improvements for persistent homology +-------------------------------------------------------------------- + +- GUDHI C++ components have been updated to the state of GUDHI v3.3.0, yielding performance improvements in ``SparseRipsPersistence``, ``EuclideanCechPersistence`` and ``CubicalPersistence`` (`#468 `_). +- Bindings for GUDHI's `edge collapser `_ have been created and can now be used as an optional preprocessing step via the optional keyword argument ``collapse_edges`` in ``VietorisRipsPersistence`` and in ``gtda.externals.ripser`` (`#469 `_ and `#483 `_). When ``collapse_edges=True``, and the input data and/or number of required homology dimensions is sufficiently large, the resulting runtimes for Vietoris–Rips persistent homology are state of the art. +- The performance of the Ripser bindings has otherwise been improved by avoiding unnecessary data copies, better managing the memory, and using more efficient matrix routines (`#501 `_ and `#507 `_). + +New transformers and functionality in ``gtda.homology`` +------------------------------------------------------- + +- The ``WeakAlphaPersistence`` transformer has been added to ``gtda.homology`` (`#464 `_). Like ``VietorisRipsPersistence``, ``SparseRipsPersistence`` and ``EuclideanCechPersistence``, it computes persistent homology from point clouds, but its runtime can scale much better with size in low dimensions. +- ``VietorisRipsPersistence`` now accepts sparse input when ``metric="precomputed"`` (`#424 `_). +- ``CubicalPersistence`` now accepts lists of 2D arrays (`#503 `_). +- A ``reduced_homology`` parameter has been added to all persistent homology transformers. When ``True``, one infinite bar in the H0 barcode is removed for the user automatically. Previously, it was not possible to *keep* these bars in the simplicial homology transformers. The default is always ``True``, which implies a breaking change in the case of ``CubicalPersistence`` (`#467 `_). + +Persistence diagrams +-------------------- + +- A ``ComplexPolynomial`` feature extraction transformer has been added (`#479 `_). +- A ``NumberOfPoints`` feature extraction transformer has been added (`#496 `_). +- An option to normalize the entropy in ``PersistenceEntropy`` according to a heuristic has been added, and a ``nan_fill_value`` parameter allows to replace any NaN produced by the entropy calculation with a fixed constant (`#450 `_). +- The computations in ``HeatKernel``, ``PersistenceImage`` and in the pairwise distances and amplitudes related to them has been changed to yield the continuum limit when ``n_bins`` tends to infinity; ``sigma`` is now measured in the same units as the filtration parameter and defaults to 0.1 (`#454 `_). + +New ``curves`` subpackage +------------------------- + +A new ``curves`` subpackage has been added to preprocess, and extract features from, collections of multi-channel curves such as returned by ``BettiCurve``, ``PersistenceLandscape`` and ``Silhouette`` (`#480 `_). It contains: + +- A ``StandardFeatures`` transformer that can extract features channel-wise in a generic way. +- A ``Derivative`` transformer that computes channel-wise derivatives of any order by discrete differences (`#492 `_). + +New ``metaestimators`` subpackage +--------------------------------- + +A new ``metaestimator`` subpackage has been added with a ``CollectionTransformer`` meta-estimator which converts any transformer instance into a fit-transformer acting on collections (`#495 `_). + +Images +------ + +- A ``DensityFiltration`` for collections of binary images has been added (`#473 `_). +- ``Padder`` and ``Inverter`` have been extended to greyscale images (`#489 `_). + +Time series +----------- + +- ``TakensEmbedding`` is now a new transformer acting on collections of time series (`#460 `_). +- The former ``TakensEmbedding`` acting on a single time series has been renamed to ``SingleTakensEmbedding`` transformer, and the internal logic employed in its ``fit`` for computing optimal hyperparameters is now available via a ``takens_embedding_optimal_parameters`` convenience function (`#460 `_). +- The ``_slice_windows`` method of ``SlidingWindow`` has been made public and renamed into ``slice_windows`` (`#460 `_). + +Graphs +------ + +- ``GraphGeodesicDistance`` has been improved as follows (`#422 `_): + + - The new parameters ``directed``, ``unweighted`` and ``method`` have been added. + - The rules on the role of zero entries, infinity entries, and non-stored values have been made clearer. + - Masked arrays are now supported. + +- A ``mode`` parameter has been added to ``KNeighborsGraph``; as in ``scikit-learn``, it can be set to either ``"distance"`` or ``"connectivity"`` (`#478 `_). + +- List input is now accepted by all transformers in ``gtda.graphs``, and outputs are consistently either lists or 3D arrays (`#478 `_). + +- Sparse matrices returned by ``KNeighborsGraph`` and ``TransitionGraph`` now have int dtype (0-1 adjacency matrices), and are not necessarily symmetric (`#478 `_). + +Mapper +------ + +- Pullback cover set labels and partial cluster labels have been added to Mapper node hovertexts (`#445 `_). + +- The functionality of ``Nerve`` and ``make_mapper_pipeline`` has been greatly extended (`#447 `_ and `#456 `_): + + - Node and edge metadata are now accessible in output ``igraph.Graph`` objects by means of the ``VertexSeq`` and ``EdgeSeq`` attributes ``vs`` and ``es`` (respectively). Graph-level dictionaries are no longer used. + - Available node metadata can be accessed by ``graph.vs[attr_name]`` where for ``attr_name`` is one of ``"pullback_set_label"``, ``"partial_cluster_label"``, or ``"node_elements"``. + - Sizes of intersections are automatically stored as edge weights, accessible by ``graph.es["weight"]``. + - A ``"store_intersections"`` keyword argument has been added to ``Nerve`` and ``make_mapper_pipeline`` to allow to store the indices defining node intersections as edge attributes, accessible via ``graph.es["edge_elements"]``. + - A ``contract_nodes`` optional parameter has been added to both ``Nerve`` and ``make_mapper_pipeline``; nodes which are subsets of other nodes are thrown away from the graph when this parameter is set to ``True``. + - A ``graph_`` attribute is stored during ``Nerve.fit``. + +- Two of the ``Nerve`` parameters (``min_intersection`` and the new ``contract_nodes``) are now available in the widgets generated by ``plot_interactive_mapper_graph``, and the layout of these widgets has been improved (`#456 `_). + +- ``ParallelClustering`` and ``Nerve`` have been exposed in the documentation and in ``gtda.mapper``'s ``__init__`` (`#447 `_). + +Plotting +-------- + +- A ``plot_params`` kwarg is available in plotting functions and methods throughout to allow user customisability of output figures. The user must pass a dictionary with keys ``"layout"`` and/or ``"trace"`` (or ``"traces"`` in some cases) (`#441 `_). +- Several plots produced by ``plot`` class methods now have default titles (`#453 `_). +- Infinite deaths are now plotted by ``plot_diagrams`` (`#461 `_). +- Possible multiplicities of persistence pairs in persistence diagram plots are now indicated in the hovertext (`#454 `_). +- ``plot_heatmap`` now accepts boolean array input (`#444 `_). + +New tutorials and examples +-------------------------- + +The following new tutorials have been added: + +- `Topology of time series `_, which explains the theory of the Takens time-delay embedding and its use with persistent homology, demonstrates the new ``API`` of several components in ``gtda.time_series``, and shows how to construct time series *classification* pipelines in ``giotto-tda`` by partially reproducing `arXiv:1910:08245 `_. +- `Topology in time series forecasting `_, which explains how to set up time series *forecasting* pipelines in ``giotto-tda`` via ``TransformerResamplerMixin``s and the ``giotto-tda`` ``Pipeline`` class. +- `Topological feature extraction from graphs `_, which explains what the features extracted from directed or undirected graphs by ``VietorisRipsPersistence``, ``SparseRipsPersistence`` and ``FlagserPersistence`` are. +- `Classifying handwritten digits `_, which presents a fully-fledged machine learning pipeline in which cubical persistent homology is applied to the classification of handwritten images from he MNIST dataset, partially reproducing `arXiv:1910.08345 `_. + +Utils +----- + +- A ``check_collection`` input validation function has been added (`#491 `_). +- ``validate_params`` now accepts ``"in"`` and ``"of"`` keys simultaneously in the ``references`` dictionaries, with ``"in"`` used for non-list-like types and ``"of"`` otherwise (`#502 `_). + +Installation improvements +------------------------- + +- ``pybind11`` is now treated as a standard git submodule in the developer installation (`#459 `_). +- ``pandas`` is now part of the testing requirements when intalling from source (`#508 `_). + +Bug Fixes +========= + +- A bug has been fixed which could lead to features with negative lifetime in persistent homology transformers when ``infinity_values`` was set too low (`#339 `_). +- By relying on ``scipy``'s ``shortest_path`` instead of ``scikit-learn``'s ``graph_shortest_path``, some errors in computing ``GraphGeodesicDistance`` (e.g. when som edges are zero) have been fixed (`#422 `_). +- A bug in the handling of COO matrices by the ``ripser`` interface has been fixed (`#465 `_). +- A bug which led to the incorrect handling of the ``homology_dimensions`` parameter in ``Filtering`` has been fixed (`#439 `_). +- An issue with the use of ``joblib.Parallel``, which led to errors when attempting to run ``HeatKernel``, ``PersistenceImage``, and the corresponding amplitudes and distances on large datasets, has been fixed (`#428 `_ and `#481 `_). +- A bug leading to plots of persistence diagrams not showing points with negative births or deaths has been fixed, as has a bug with the computation of the range to be shown in the plot (`#437 `_). +- A bug in the handling of persistence pairs with negative death values by ``Filtering`` has been fixed (`#436 `_). +- A bug in the handling of ``homology_dimension_ix`` (now renamed to ``homology_dimension_idx``) in the ``plot`` methods of ``HeatKernel`` and ``PersistenceImage`` has been fixed (`#452 `_). +- A bug in the labelling of axes in ``HeatKernel`` and ``PersistenceImage`` plots has ben fixed (`#453 `_ and `#454 `_). +- ``PersistenceLandscape`` plots now show all homology dimensions, instead of just the first (`#454 `_). +- A bug in the computation of amplitudes and pairwise distances based on persistence images has been fixed (`#454 `_). +- ``Silhouette`` now does not create NaNs when a subdiagram is trivial (`#454 `_). +- ``CubicalPersistence`` now does not create pairs with negative persistence when ``infinity_values`` is set too low (`#467 `_). +- Warnings are no longer thrown by ``KNeighborsGraph`` when ``metric="precomputed"`` (`#506 `_). +- A bug in ``Labeller.resample`` affecting cases in which ``n_steps_future >= size - 1``, has been fixed (`#460 `_). +- A bug in ``validate_params``, affecting the case of tuples of allowed types, has been fixed (`#502 `_). + +Backwards-Incompatible Changes +============================== + +- The minimum required versions from most of the dependencies have been bumped. The updated dependencies are ``numpy >= 1.19.1``, ``scipy >= 1.5.0``, ``joblib >= 0.16.0``, ``scikit-learn >= 0.23.1``, ``python-igraph >= 0.8.2``, ``plotly >= 4.8.2``, and ``pyflagser >= 0.4.1`` (`#457 `_). +- ``GraphGeodesicDistance`` now returns either lists or 3D dense ndarrays for compatibility with the homology transformers - By relying on ``scipy``'s ``shortest_path`` instead of ``scikit-learn``'s ``graph_shortest_path``, some errors in computing ``GraphGeodesicDistance`` (e.g. when som edges are zero) have been fixed (`#422 `_). +- The output of ``PairwiseDistance`` has been transposed to match ``scikit-learn`` convention ``(n_samples_transform, n_samples_fit)`` (`#420 `_). +- ``plot`` class methods now return figures instead of showing them (`#441 `_). +- Mapper node and edge attributes are no longer stored as graph-level dictionaries, ``"node_id"`` is no longer an available node attribute, and the attributes ``nodes_`` and ``edges_`` previously stored by ``Nerve.fit`` have been removed in favour of a ``graph_`` attribute (`#447 `_). +- The ``homology_dimension_ix`` parameter available in some transformers in ``gtda.diagrams`` has been renamed to ``homology_dimensions_idx`` (`#452 `_). +- The base of the logarithm used by ``PersistenceEntropy`` is now 2 instead of *e*, and NaN values are replaced with -1 instead of 0 by default (`#450 `_ and `#474 `_). +- The outputs of ``PersistenceImage``, ``HeatKernel`` and of the pairwise distances and amplitudes based on them is now different due to the improvements described above. +- Weights are no longer stored in the ``effective_metric_params_`` attribute of ``PairwiseDistance``, ``Amplitude`` and ``Scaler`` objects when the metric is persistence-image–based; only the weight function is (`#454 `_). +- The ``homology_dimensions_`` attributes of several transformers have been converted from lists to tuples. When possible, homology dimensions stored as parts of attributes are now presented as ints (`#454 `_). +- ``gaussian_filter`` (used to make heat– and persistence-image–based representations/pairwise distances/amplitudes) is now called with ``mode="constant"`` instead of ``"reflect"`` (`#454 `_). +- The default value of ``order`` in ``Amplitude`` has been changed from ``2.`` to ``None``, giving vector instead of scalar features (`#454 `_). +- The meaning of the default ``None`` for ``weight_function`` in ``PersistenceImage`` (and in ``Amplitude`` and ``PairwiseDistance`` when ``metric="persistence_image"``) has been changed from the identity function to the function returning a vector of ones (`#454 `_). +- Due to the updates in the GUDHI components, some of the bindings and Python interfaces to the GUDHI C++ components in ``gtda.externals`` have changed (`#468 `_). +- ``Labeller.transform`` now returns a 1D array instead of a column array (`#475 `_). +- ``PersistenceLandscape`` now returns 3D arrays instead of 4D ones, for compatibility with the new ``curves`` subpackage (`#480 `_). +- By default, ``CubicalPersistence`` now removes one infinite bar in H0 (`#467 `_, and see above). +- The former ``width`` parameter in ``SlidingWindow`` and ``Labeller`` has been replaced with a more intuitive ``size`` parameter. The relation between the two is: ``size = width + 1`` (`#460 `_). +- ``clusterer`` is now a required parameter in ``ParallelClustering`` (`#508 `_). +- The ``max_fraction`` parameter in ``FirstSimpleGap`` and ``FirstHistogramGap`` now indicates the floor of ``max_fraction * n_samples``; its default value has been changed from ``None`` to ``1`` (`#412 `_). + +Thanks to our Contributors +========================== + +This release contains contributions from many people: + +Umberto Lupo, Guillaume Tauzin, Julian Burella Pérez, Wojciech Reise, Lewis Tunstall, Nick Sale, and Anibal Medina-Mardones. + +We are also grateful to all who filed issues or helped resolve them, asked and answered questions, and were part of inspiring discussions. + ************* Release 0.2.2 ************* @@ -439,4 +620,4 @@ for this implementation. Release 0.1a.0 ************** -Initial release of the library, original named ``giotto-learn``. +Initial release of the library, originally named ``giotto-learn``. diff --git a/examples/MNIST_classification.ipynb b/examples/MNIST_classification.ipynb index 59f86de39..34c130b83 100644 --- a/examples/MNIST_classification.ipynb +++ b/examples/MNIST_classification.ipynb @@ -90,7 +90,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Create train and test sets" + "### Create train and test sets" ] }, { @@ -149,7 +149,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "# Binarize the image" + "### Binarize the image" ] }, { @@ -491,7 +491,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "### Training a classifier" + "## Training a classifier" ] }, { diff --git a/examples/persistent_homology_graphs.ipynb b/examples/persistent_homology_graphs.ipynb index a6a91c76a..224b76586 100644 --- a/examples/persistent_homology_graphs.ipynb +++ b/examples/persistent_homology_graphs.ipynb @@ -130,6 +130,7 @@ "To understand what these persistence diagrams are telling us about the input weighted graphs, we briefly explain the **clique complex (or flag complex) filtration** procedure underlying the computations in ``VietorisRipsPersistence`` when ``metric=\"precomputed\"``, via an example.\n", "\n", "Let us start with a special case of a weighted graph with adjacency matrix as follows:\n", + "\n", "- the diagonal entries (\"vertex weights\") are all zero;\n", "- all off-diagonal entries (edge weights) are non-negative;\n", "- some edge weights are infinite (or very very large).\n", @@ -141,6 +142,7 @@ "The procedure can be explained as follows: we let a parameter $\\varepsilon$ start at 0, and as we increase it all the way to infinity we keep considering the instantaneous subgraphs made of a) all the vertices in the original graph, and b) those edges whose weight is less than or equal to the current $\\varepsilon$. We also promote these subgraphs to more general structures called **(simplicial) complexes** that, alongside vertices and edges, also possess $k$**-simplices**, i.e. selected subsets of $k + 1$ vertices (a 2-simplex is an abstract \"triangle\", a 3-simplex an abstract \"tetrahedron\", etc). Our criterion is this: for each integer $k \\geq 2$, all $(k + 1)$-cliques in each instantaneous subgraph are declared to be the $k$-simplices of the subgraph's associated complex. By definition, the $0$-simplices are the vertices and the $1$-simplices are the available edges.\n", "\n", "As $\\varepsilon$ increases from 0 (included) to infinity, we record the following information:\n", + "\n", "1. How many new **connected components** are created because of the appearance of vertices (in this example, all vertices \"appear\" in one go at $\\varepsilon = 0$, by definition!), or merge because of the appearance of new edges.\n", "2. How many new 1-dimensional \"holes\", 2-dimensional \"cavities\", or more generally $d$-dimensional **voids** are created in the instantaneous complex. A hole, cavity, or $d$-dimensional void is such only if there is no collection of \"triangles\", \"tetrahedra\", or $(d + 1)$-simplices which the void is the \"boundary\" of. *Note*: Although the edges of a triangle *alone* \"surround a hole\", these cannot occur in our particular construction because the \"filling\" triangle is also declared present in the complex when all its edges are.\n", "3. How many $d$-dimensional voids which were present at earlier values of $\\epsilon$ are \"filled\" by $(d + 1)$-simplices which just appear.\n", @@ -242,6 +244,7 @@ "And just as in the case of weighted graphs, we record the appearance/disappearance of connected components and voids as we keep increasing $r$.\n", "\n", "The case of point clouds can actually be thought of as a special case of the case of FCW graphs. Namely, if:\n", + "\n", "1. we regard each point in the cloud as an abstract vertex in a graph,\n", "2. we compute the square matrix of pairwise (Euclidean or other) distances between points in the cloud, and\n", "3. we run the procedure explained above with $\\varepsilon$ defined as $2r$,\n", @@ -278,8 +281,9 @@ "What if, as is the case in many applications, our graphs have sparse connections and are unweighted?\n", "\n", "In ``giotto-tda``, there are two possibilities:\n", - "1. One can encode the graphs as adjacency matrices of non-fully connected weighted graphs, where all weights corresponding to edges which are present are equal to ``1.`` (or any other positive constant). See section ***Non-fully connected weighted graphs*** above for the different encoding conventions for sparse and dense matrices.\n", - "2. One can preprocess the unweighted graph via [GraphGeodesicDistance](https://giotto-ai.github.io/gtda-docs/latest/modules/generated/graphs/processing/gtda.graphs.GraphGeodesicDistance.html) to obtain a FCW graph where edge $ij$ has as weight the length of the shortest path from vertex $i$ to vertex $j$ (and ``np.inf`` if no path exists between the two vertices in the original graph).\n", + "\n", + "1. Encode the graphs as adjacency matrices of non-fully connected weighted graphs, where all weights corresponding to edges which are present are equal to ``1.`` (or any other positive constant). See section ***Non-fully connected weighted graphs*** above for the different encoding conventions for sparse and dense matrices.\n", + "2. Preprocess the unweighted graph via [GraphGeodesicDistance](https://giotto-ai.github.io/gtda-docs/latest/modules/generated/graphs/processing/gtda.graphs.GraphGeodesicDistance.html) to obtain a FCW graph where edge $ij$ has as weight the length of the shortest path from vertex $i$ to vertex $j$ (and ``np.inf`` if no path exists between the two vertices in the original graph).\n", "\n", "### Example 1: Circle graph\n", "\n", diff --git a/examples/vietoris_rips_quickstart.ipynb b/examples/vietoris_rips_quickstart.ipynb index 0b89b0a0a..0f7431a88 100644 --- a/examples/vietoris_rips_quickstart.ipynb +++ b/examples/vietoris_rips_quickstart.ipynb @@ -62,7 +62,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "**Important note**: ``VietorisRipsPersistence``, and all other \"persistence homology\" transformers in ``gtda.homology``, expect input in the form of a 3D array or, in some cases, a list of 2D arrays. For each entry in the input (here, for each point cloud in ``point_clouds``) they compute a topological summary which is also a 2D array, and then stack all these summaries into a single output 3D array. So, in our case, ``diagrams[i]`` represents the topology of ``point_clouds[i]``. ``diagrams[i]`` is interpreted as follows:\n", + "**Important note**: ``VietorisRipsPersistence``, and all other \"persistent homology\" transformers in ``gtda.homology``, expect input in the form of a 3D array or, in some cases, a list of 2D arrays. For each entry in the input (here, for each point cloud in ``point_clouds``) they compute a topological summary which is also a 2D array, and then stack all these summaries into a single output 3D array. So, in our case, ``diagrams[i]`` represents the topology of ``point_clouds[i]``. ``diagrams[i]`` is interpreted as follows:\n", "- Each row is a triplet describing a single topological feature found in ``point_clouds[i]``.\n", "- The first and second entries (respectively) in the triplet denote the values of the \"filtration parameter\" at which the feature appears or disappears respectively. They are referred to as the \"birth\" and \"death\" values of the feature (respectively). The meaning of \"filtration parameter\" depends on the specific transformer, but in the case of ``VietorisRipsPersistence`` on point clouds it has the interpretation of a length scale.\n", "- A topological feature can be a connected component, 1D hole/loop, 2D cavity, or more generally $d$-dimensional \"void\" which exists in the data at scales between its birth and death values. The integer $d$ is the *homology dimension* (or degree) of the feature and is stored as the third entry in the triplet. In this example, the shapes should have 2D cavities so we explicitly tell ``VietorisRipsPersistence`` to look for these by using the ``homology_dimensions`` parameter!\n", diff --git a/gtda/_version.py b/gtda/_version.py index 589f7abfa..b6700d318 100644 --- a/gtda/_version.py +++ b/gtda/_version.py @@ -19,4 +19,4 @@ # 'X.Y.dev0' is the canonical version of 'X.Y.dev' # -__version__ = '0.2.2' +__version__ = "0.3.0" diff --git a/gtda/diagrams/preprocessing.py b/gtda/diagrams/preprocessing.py index a036ea2bd..1f1d0464e 100644 --- a/gtda/diagrams/preprocessing.py +++ b/gtda/diagrams/preprocessing.py @@ -471,7 +471,7 @@ def transform(self, X, y=None): Xt : ndarray of shape (n_samples, n_features_filtered, 3) Filtered persistence diagrams. Only the subdiagrams corresponding to dimensions in :attr:`homology_dimensions_` are filtered. - ``n_features_filtered`` is less than or equal to ``n_features`. + ``n_features_filtered`` is less than or equal to ``n_features``. """ check_is_fitted(self) diff --git a/setup.py b/setup.py index f082312cd..3c76b4075 100755 --- a/setup.py +++ b/setup.py @@ -29,7 +29,7 @@ MAINTAINER_EMAIL = "maintainers@giotto.ai" URL = "https://github.com/giotto-ai/giotto-tda" LICENSE = "GNU AGPLv3" -DOWNLOAD_URL = "https://github.com/giotto-ai/giotto-tda/tarball/v0.2.2" +DOWNLOAD_URL = "https://github.com/giotto-ai/giotto-tda/tarball/v0.3.0" VERSION = __version__ # noqa CLASSIFIERS = ["Intended Audience :: Science/Research", "Intended Audience :: Developers",