Bug in saving HDF5 file with None values

[gretatuckute]

Hi, I encountered a bug when saving HDF5 files. I am running GLMsingle in Python without standard cross-validation (i.e., a fixed pcstop and fracs value). I am using the most recent version of GLMsingle.

Here is the error message:

Traceback (most recent call last):
  File "evaluate_GLMsingle_focal.py", line 259, in <module>
    main()
  File "evaluate_GLMsingle_focal.py", line 237, in main
    results_glmsingle = glmsingle_obj.fit(
  File "/om2/user/gretatu/anaconda/envs/glmsinglev2/lib/python3.8/site-packages/glmsingle/glmsingle.py", line 1711, in fit
    hf.create_dataset(k, data=v)
  File "/om2/user/gretatu/anaconda/envs/glmsinglev2/lib/python3.8/site-packages/h5py/_hl/group.py", line 183, in create_dataset
    dsid = dataset.make_new_dset(group, shape, dtype, data, name, **kwds)
  File "/om2/user/gretatu/anaconda/envs/glmsinglev2/lib/python3.8/site-packages/h5py/_hl/dataset.py", line 52, in make_new_dset
    raise TypeError("One of data, shape or dtype must be specified")
TypeError: One of data, shape or dtype must be specified

I fixed it via the PR (not saving values that are None). I also added some print statements in my code, and these values seemed problematic:

Warning: Skipping key 'glmbadness' because its value is None or unsupported: None
Warning: Skipping key 'pcvoxels' because its value is None or unsupported: None
Warning: Skipping key 'xvaltrend' because its value is None or unsupported: None
Warning: Converting list to numpy array for key 'pcregressors', shape of values: (8, 170, 11)
Warning: Converting list to numpy array for key 'rrbadness', shape of values: (0,)

(should I be worried? Is this unexpected, or simply a bug in saving values that do not exist in the non-cross-validated use case?)

I ran:

results_glmsingle = glmsingle_obj.fit( design, data, stimdur, tr, outputdir=OUTPUTDIR)

where design is a list of length 8, with numpy arrays of shape (170; 400), data is a list of length 8 with numpy arrays of shape (91; 109; 91; 170), stimdur is 2 and tr is 2.

The opt params are:

{'R2thresh': 0, 'brainR2': [], 'brainexclude': False, 'brainthresh': [99.0, 0.1], 'chunklen': 50000, 'extra_regressors': False, 'fracs': 0.05, 'hrffitmask': 1, 'hrfmodel': 'optimise', 'hrfthresh': 0.5, 'lambda': 0, 'n_boots': 100, 'n_jobs': 1, 'n_pcs': 10, 'numforhrf': 50, 'pcR2cutoff': [], 'pcR2cutoffmask': 1, 'pcstop': -5, 'seed': 1708355214.6366038, 'suppressoutput': 0, 'wantautoscale': 1, 'wantfileoutputs': [1, 1, 1, 1], 'wantfracridge': 1, 'wantglmdenoise': 1, 'wanthdf5': 1, 'wantlibrary': 1, 'wantlss': 0, 'wantmemoryoutputs': [1, 1, 1, 1], 'wantparametric': 0, 'wantpercentbold': 1}

Thanks a lot :)