MRG: implement ResidualizeTarget #11
Conversation
|
You also need to few tests corresponding to this class! |
|
|
|
||
| in_targets = check_array(in_targets) | ||
| confounds = check_array(confounds, ensure_2d=False) | ||
|
|
There was a problem hiding this comment.
Add checks for data type to ensure it is numeric
|
|
||
| check_is_fitted(self, 'model_', 'n_targets_') | ||
| test_targets = check_array(test_targets, accept_sparse=True) | ||
|
|
confounds/base.py
Outdated
| Parameters | ||
| ---------- | ||
| X : {array-like, sparse matrix}, shape (n_samples, n_targets) | ||
| The training input samples. |
There was a problem hiding this comment.
make it clear X is targets here, although it is called X
| max_dim = 100 | ||
| for n_samples in np.random.randint(0, 20, 1): | ||
| for num_confounds in np.random.randint(min_dim, max_dim, 3): | ||
| train_all, _ = make_sparse_uncorrelated( |
There was a problem hiding this comment.
this does not generate discrete numerical/integer values! We need a way to broaden the test with multiple data types!
There was a problem hiding this comment.
a bit confused here, does this look better? np.random.randint(0, 20, 1, dtype=int)
There was a problem hiding this comment.
before updating the tests, how about you try making a jupyter notebook, and play with this class, and write some examples, to see how it works?
There was a problem hiding this comment.
hey @raamana sorry for the late response. this summer has been a bit crazy. probably need more time-out here. i noticed that some repos have been snapshotted in Arctic Code Vault, i guess it didn't snapshot PR :p
There was a problem hiding this comment.
NP, take your time. Happy to chat with you to discuss this and make a more precise plan to finish it quickly.
why are you referencing the Arctic Code Vault? did you lose your fork or what?
There was a problem hiding this comment.
thanks! i'm wondering if the snapshot in Arctic Code Vault would include PR? and no, i didn't lose forks.
There was a problem hiding this comment.
I don't know - but I am sure your contribution will be clearly recognized :)
There was a problem hiding this comment.
hi @zuxfoucault, would you have time this year to work on this?
implement ResidualizeTarget at HBM brainhack