0.2.1

Bugfix on the repr utility function.

What's Changed

• Hotfix repr by @BalzaniEdoardo in #296

Full Changelog: 0.2.0...0.2.1

0.2.0

Overview

• New basis API. Distinct basis objects for "Conv" and "Eval" mode of operation.
• Basis compute_features accepts multi-dimensional inputs for both "Conv" and "Eval" bases.
• Added IdentityEval bases for including custom inputs in composite bases.
• Added HistoryConv bases to include the raw history of an input as a predictor.
• New module for applying identifiability constraints to a model matrix.
• New SVRG solver with optimized parameters.
• New documentation website based on sphinx PyData theme.
• Improved basis documentation: new how-to guides and summary of the available bases.
• Added examples in docstrings of user-facing methods and functions.
• Coefficient splitting method of basis.

What's Changed

• Update python versions by @BalzaniEdoardo in #211
• Add (Prox-)SVRG as a solver for GLMs by @bagibence in #184
• fix ident constraint by @BalzaniEdoardo in #214
• Swap parameter order to match sklearn scorer by @BalzaniEdoardo in #215
• Use pooch by @BalzaniEdoardo in #216
• add try/except for seaborn and ipython by @BalzaniEdoardo in #218
• improved code refs by @BalzaniEdoardo in #219
• Change names and simplify landing page by @BalzaniEdoardo in #223
• set up javascript animation by @BalzaniEdoardo in #230
• renaming files in dev notes by @BalzaniEdoardo in #231
• Development by @BalzaniEdoardo in #233
• Add support section by @BalzaniEdoardo in #234
• Doctest by @BalzaniEdoardo in #232
• reordered cards by @BalzaniEdoardo in #237
• Move overview by @BalzaniEdoardo in #227
• prevent absolute links to docs (except badge) by @BalzaniEdoardo in #238
• Fix sklearn test by @BalzaniEdoardo in #240
• Bugfix basis properties by @BalzaniEdoardo in #241
• Revert "Bugfix basis properties" by @BalzaniEdoardo in #244
• Bugfix basis properties by @BalzaniEdoardo in #245
• fixed warns in test_convergence.py by @BalzaniEdoardo in #242
• Fix warns test convolution by @BalzaniEdoardo in #246
• Development by @BalzaniEdoardo in #239
• Splits tox into multiple jobs by @billbrod in #250
• require sample axis to be axis=0 by @BalzaniEdoardo in #254
• Initialization by @BalzaniEdoardo in #252
• Added glm.py examples in docstrings by @pranmod01 in #249
• Automatic step sizes for SVRG by @bagibence in #207
• added linting tests to tox.ini and linted by @BalzaniEdoardo in #258
• Started basis examples by @pranmod01 in #253
• fixed abs links by @BalzaniEdoardo in #262
• Start facilitating coefficient splitting by @BalzaniEdoardo in #247
• Sphinx by @BalzaniEdoardo in #264
• Basis refactor pr1 by @BalzaniEdoardo in #273
• Document basis by @BalzaniEdoardo in #274
• Multi dim eval basis by @BalzaniEdoardo in #275
• Set shape basis method by @BalzaniEdoardo in #276
• Improve transformer api by @BalzaniEdoardo in #280
• Identity basis by @BalzaniEdoardo in #282
• added decorator to simulate by @BalzaniEdoardo in #287
• PopulationGLM sklearn clone by @billbrod in #291
• fix split_by_feature bug by @billbrod in #289
• Adds repr for our custom classes by @billbrod in #286
• Development by @BalzaniEdoardo in #293
• Bugfix init composite basis by @BalzaniEdoardo in #294
• New release by @BalzaniEdoardo in #295

New Contributors

• @pranmod01 made their first contribution in #249

Full Changelog: 0.1.6...0.2.0

0.1.6

NeMoS New Release

In this new release:

• Simplified API for GLM: You can now set regularization using a string. All regularization and solver hyperparameters can be passed during GLM initialization.

  import nemos as nmo
  model = nmo.glm.GLM(regularizer="Ridge", regularizer_strength=0.1)

• New TransformerBasis API: All Basis attributes are now available in TransformerBasis, facilitating inspection and cross-validation. See the updated tutorial in the API guide.

  import nemos as nmo
  transformer_basis = nmo.basis.RaisedCosineBasisLinear(10).to_transformer()
  print(transformer_basis.n_basis_funcs)  # prints 10

• Bounds for Basis: Added support for specifying bounds for the range of values covered by a basis function.

What's Changed

• quickfix by @BalzaniEdoardo in #175
• improved docstrings by @BalzaniEdoardo in #182
• run ci only for non-draft prs by @BalzaniEdoardo in #192
• Refactoring regularizer and solver by @clewis7 in #183
• add tests for checking get_params() by @clewis7 in #200
• Provide a range for basis by @BalzaniEdoardo in #191
• update contributing guide by @clewis7 in #198
• fix convergence tests by @clewis7 in #206
• Update TransformerBasis and Basis to work with scikit-learn pipelining and CV by @bagibence in #169
• fixed basis call docstrings by @BalzaniEdoardo in #209

New Contributors

• @clewis7 made their first contribution in #183 🚀 🚀 🚀
• @bagibence made their first contribution in #169 🚀 🚀 🚀

Full Changelog: 0.1.5...0.1.6

NeMoS version 0.1.5

New with this release:

• `GammaObservations model, for fitting Gamma-GLMs.
• GLMs' update and initialize_solver methods for batching

NeMoS version 0.1.4

New with this release:

• GammaObservations model, for fitting Gamma-GLMs.

• GLMs' update and initialize_solver methods for batching

Compatiblity with `pynapple>=0.6.5`

This update make sure that nemos is compatible with pynapple>=0.6.5.

`nemos` for GLM and `PopulationGLM`

Release of the package tested for python 3.9, 3.10, and 3.11. Current modules include basis.py for feature construction and glm for single neuron and population GLM modeling.

`nemos` first release

First working release of nemos.

Reserve package name on PyPI

This is an alpha version of the code. We are releasing this version with the aim of reserving the package name on PyPI