question on groupby.agg

[Neal Becker]

Admittedly, my use my seem odd.  I have used groupby.agg to combine results from multiple experiments, as we discussed a couple of weeks ago.  But now I have a new problem.  I have data where elements are not scalar but vectors.  So I have a dataframe where some columns are a vector (using a list or a tuple).

If I try to use groupby.agg on this, I get:

  File "./plot_stuff2.py", line 231, in combined_cols

    rdf = df.groupby(list(columns_to_group)).agg(d).reset_index()

  File "/home/nbecker/.local/lib/python2.7/site-packages/pandas/core/groupby.py", line 344, in agg

    return self.aggregate(func, *args, **kwargs)

  File "/home/nbecker/.local/lib/python2.7/site-packages/pandas/core/groupby.py", line 1727, in aggregate

    result[col] = colg.aggregate(agg_how)

  File "/home/nbecker/.local/lib/python2.7/site-packages/pandas/core/groupby.py", line 1420, in aggregate

    return self._python_agg_general(func_or_funcs, *args, **kwargs)

  File "/home/nbecker/.local/lib/python2.7/site-packages/pandas/core/groupby.py", line 476, in _python_agg_general

    result, counts = self.grouper.agg_series(obj, f)

  File "/home/nbecker/.local/lib/python2.7/site-packages/pandas/core/groupby.py", line 883, in agg_series

    return self._aggregate_series_pure_python(obj, func)

  File "/home/nbecker/.local/lib/python2.7/site-packages/pandas/core/groupby.py", line 917, in _aggregate_series_pure_python

    raise ValueError('Function does not reduce')

ValueError: Function does not reduce

I wonder if this restriction is unnecessary?  Or maybe I just shouldn't be trying to use groupby.agg for this.  Or perhaps I should be using some other approach entirely.

The purpose of this is to combine results from experiments.  The combinations would be simple functions like sum.  In this case, it would be sum element-by-element of these vectors, producing a vector result to be put into the new df.

That is, I might have a df like:

0 (1,2,3)

1 (4,5,6)

And the reduction should produce

0 (5, 7, 9)

[Wouter Overmeire]

2013/11/25 Neal Becker <ndbe...@gmail.com>

--
You received this message because you are subscribed to the Google Groups "PyData" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pydata+un...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

Does this example help?

 

In [151]: df

Out[151]: 

              A  B

0     (1, 2, 3)  0

1     (4, 5, 6)  0

2     (7, 8, 9)  1

3  (10, 11, 12)  1

In [152]: myagg = lambda s:tuple(sum(i) for i in zip(*s))                                                                                                                                                                                                                                                                      

In [153]: df.groupby('B').agg(myagg)

Out[153]: 

              A

B              

0     (5, 7, 9)

1  (17, 19, 21)