cophi-wue / pydelta Goto Github PK

When doing

delta.graphics.spikeplot(corpus, docs = [2,5])

I get an

AttributeError: 'Corpus' object has no attribute 'ix'

This seems related to the depreciation of .ix() in pandas, see https://stackoverflow.com/questions/59991397/attributeerror-dataframe-object-has-no-attribute-ix or https://stackoverflow.com/questions/43838999/pandas-replacement-for-ix

Since (I guess?) 0.16, pandas supports a _constructor property on DataFrame or Series subclasses that will be used when pandas needs to create a new version of one of its datatypes. This could be used to keep track of our metadata in an easier and more robust way than the current manual implementation, where, e.g. corpus / scalar will always create a plain DataFrame and we have to copy the metadata manually, by implementing something like:

class Corpus:

     _metadata = ['metadata']

     @property
     def _constructor(self):
         def constructor(*args, **kwargs):
	     result = Corpus(*args, **kwargs)
	     result.metadata.update_from(self)
	     return result
	 return constructor

     ...

It would require to get rid of much of the conversion and copying code, though.

In deltas.py: dm += dm.T is for some reason not the same as dm = dm + dm.T
Fix:
next...andbue:next#diff-f46c297a1abc9e356cb9e07b2a4d5938

I tried the pip command with the download

pip install git+https://github.com/cophi-wue/pydelta@next

and have this error:

ERROR: Package 'delta' requires a different Python: 3.10.6 not in '<3.10,>=3.7.1'

If a new CompositeDeltaFunction is created, its fix_symmetry-attribute is for some reason set to "False".
Is this the expected behaviour?

The KMedoids implementation (that is planned to be integrated for 0.20, see sklearn issue 7694) doesn't currently compile against the pydelta/next requirements list. Since this is integrated using requirements.txt, we'll probably either need to remove it from there or update the dependency.

    distances.evaluate()
  File "/tmp/pydelta/delta/deltas.py", line 787, in evaluate
    result["Simple Score"] = self.simple_score()
  File "/tmp/pydelta/delta/deltas.py", line 772, in simple_score
    in_group_df, out_group_df = self.z_scores().partition()
  File "/tmp/pydelta/delta/deltas.py", line 731, in z_scores
    return DistanceMatrix((self - deltas.mean()) / deltas.std(),
  File "/tmp/pydelta/.venv/lib/python3.8/site-packages/pandas/core/ops/__init__.py", line 660, in f
    new_data = dispatch_to_series(self, other, op)
  File "/tmp/pydelta/.venv/lib/python3.8/site-packages/pandas/core/ops/__init__.py", line 266, in dispatch_to_series
    return type(left)(bm)
  File "/tmp/pydelta/delta/deltas.py", line 630, in __init__
    self.metadata = Metadata(metadata, **kwargs)
  File "/tmp/pydelta/delta/util.py", line 31, in __init__
    self.update(*args, **kwargs)
  File "/tmp/pydelta/delta/util.py", line 84, in update
    self._update_from(arg)
  File "/tmp/pydelta/delta/util.py", line 53, in _update_from
    self.__dict__.update(d)
TypeError: 'NoneType' object is not iterable

works for pandas < 0.25

Thomas’/Frimis Verfahren