Add option to not sort within groups in GroupBy

[wesm]

No description provided.

[stig]

I'd love an option to maintain the column order that I've specified. At the moment I do this:

    print df.groupby('Status')['Duration'].agg({ 'Min': np.min,
                                                 'Max':np.max,
                                                 'Mean': np.mean,
                                                 'Median': np.median,
                                                 'Stddev':np.std,
                                                 })

but the column order come out like this:

         Max    Mean     Median  Min    Stddev
Error    0.141  0.09967  0.093   0.072  0.02796
Success  12.07  5.879    11.05   0.27   5.41

Which is not ideal. I'd love to pass an OrderedDict and have the order be respected.

    d = OrderedDict([('Min', np.min),
                     ('Max', np.max),
                     ('Mean', np.mean),
                     ('Median', np.median),
                     ('Stddev', np.std),
                     ])

... or, the method could just accept a list of tuples and respect that order.

[wesm]

@stig i'll open a separate issue for the thing you mention which I agree with

[soupault]

@stig Hi, any news or plans on this issue?

[jreback]

pass sort=False

[soupault]

@jreback brilliant, thanks a lot!