Ronan Perry
Ronan Perry
I am unsure why, but occasionally the fisherz test experiences the following error. I have not had this problem with the KCI test on the same data. I do not...
The current documentation takes ~14 minutes to run on each PR update. I think that this is too computationally inefficient and ideally could be resolved somehow (simpler tutorials, selective updating,...
Only seems to be an issue in Python 3.8 `=========================== short test summary info ============================ FAILED tests/embed/test_mcca.py::test_initial_svds[True] - assert 3 == 4 FAILED tests/embed/test_mcca.py::test_initial_svds[False] - assert 3 == 4`
Matplotlib 3.4.0 introduced changes (at least with the number of parameters allowed in `ax.set_ticks` I believe) that are incompatible with some of the current examples, see error below. There is...
Adds an example showing CCA being used for cross-view information retrieval on the UCI dataset. We didn't have a non-simulated data example for CCA methods.
On line 346 of gcca.py, the projections are scaled by the squareroot of the number of features which leads to in sample embeddings with that norm. Should remove as is...
MV kmeans as implemented assumes that observations are clustered the same across each view. This assumes that the views contain the same information, but some applications it may be reasonable...
Enhancements and TODOs for multiview CCA - [ ] Allow alternate `fit` for precomputed SVD decomposition components. - [ ] If no regularization, check that all views are low-dimensional -...
## Expected Behavior If a soft dependency isn't installed, no error should be thrown but the corresponding tools shouldn't be imported as the dependency raises an error on import. If...
see a list of topics at https://github.com/microsoft/graspologic/issues/531