Trouble running 02_haxby_fmri: visualize stages of mapper

[jsmentch]

I can't seem to run 4.2 Visualize the stages of mapper in the haxby example. This works in the trefoil knot example as it did before.

I think there are also other things that do not work, all of the cells after that don't work as well. The outputs of kmapper seem to be very different from when the notebook was last updated. One of the lines says "Created 0 edges and 0 nodes" for example. Been trying to figure this out but haven't been able to get it to work.

This is the error I get after trying to visualize the stages of mapper:

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
/opt/miniconda-latest/envs/neuro/lib/python3.7/site-packages/dyneusr/tools/networkx_utils.py in get_cover_cubes(lens, graph, cover, scale, **kwargs)
    235         cover_cubes = {}
--> 236         for i, center in enumerate(cover.centers_):
    237             lower = center - cover.radius_

TypeError: 'NoneType' object is not iterable

During handling of the above exception, another exception occurred:

AttributeError                            Traceback (most recent call last)
<ipython-input-14-23a7a1134df5> in <module>
      5 _ = ds.tools.networkx_utils.visualize_mapper_stages(
      6     data=subject, lens=lens, cover=cover, graph=graph,dG=dG,
----> 7     node_size=10,
      8     )

/opt/miniconda-latest/envs/neuro/lib/python3.7/site-packages/dyneusr/tools/networkx_utils.py in visualize_mapper_stages(data, y, lens, cover, graph, dG, **kwargs)
    431 
    432     # 2. draw cover (axes: 2)
--> 433     draw_cover(ax=axes[1], graph=graph, lens=lens2D, cover=cover)
    434 
    435     # 3. draw clusters (axes: 3)

/opt/miniconda-latest/envs/neuro/lib/python3.7/site-packages/dyneusr/tools/networkx_utils.py in draw_cover(ax, cover_cubes, draw_all, max_draw, **kwargs)
    286 
    287     if cover_cubes is None:
--> 288         cover_cubes = get_cover_cubes(**kwargs)
    289 
    290     # extract bins (TODO: probably a better way to do this)

/opt/miniconda-latest/envs/neuro/lib/python3.7/site-packages/dyneusr/tools/networkx_utils.py in get_cover_cubes(lens, graph, cover, scale, **kwargs)
    242 
    243         # support deprecated Cover API from kmapper==1.1.6
--> 244         bins = [tuple(_) for _ in cover.define_bins(ilens)]
    245         chunk_dist = cover.chunk_dist
    246         overlap_dist = cover.overlap_dist

AttributeError: 'Cover' object has no attribute 'define_bins'

[calebgeniesse]

Sorry for the delayed response, but thanks for reporting this.

It seems like kmapper's API has diverged since the notebooks were first developed. For now, you should be able to resolve the errors and behavior using kmapper==1.1.6.

At some point, the dyneusr codebase, as well as the parameters used in these notebooks, should be updated to ensure compatibility with more recent releases of kmapper.

[calebgeniesse]

I don't have time to resolve this fully just yet, but I will add this to the top of my to-do list.

[jsmentch]

Ok thanks!
For future reference if it helps to track this down: I installed kmapper==1.1.6 but am getting a different error now, when try to import dyneusr:

---------------------------------------------------------------------------
ModuleNotFoundError                       Traceback (most recent call last)
<ipython-input-12-691af6a23625> in <module>
----> 1 import dyneusr as ds
      2 
      3 # kmapper wrappers
      4 from dyneusr.mapper.wrappers import KMapperWrapper
      5 from dyneusr.mapper.utils import optimize_cover, optimize_dbscan

/opt/miniconda-latest/envs/neuro/lib/python3.7/site-packages/dyneusr/__init__.py in <module>
     15 
     16 from . import datasets
---> 17 from . import mapper
     18 from . import tools
     19 from . import visuals

/opt/miniconda-latest/envs/neuro/lib/python3.7/site-packages/dyneusr/mapper/__init__.py in <module>
      6 
      7 from . import utils
----> 8 from . import wrappers
      9 # from . import persistence
     10 

/opt/miniconda-latest/envs/neuro/lib/python3.7/site-packages/dyneusr/mapper/wrappers.py in <module>
     12 # Machine learning libraries
     13 from sklearn.datasets.base import Bunch
---> 14 from sklearn.externals.joblib import Memory
     15 from sklearn.base import BaseEstimator, TransformerMixin, ClusterMixin
     16 from sklearn.preprocessing import MinMaxScaler

ModuleNotFoundError: No module named 'sklearn.externals.joblib'

And this is with:
dyneusr 0.3.8
kmapper 1.1.6
sklearn: 0.23.0