ENH: pct_change in a group by return groups

[marcdelabarrera]

Feature Type

• Adding new functionality to pandas

• Changing existing functionality in pandas

• Removing existing functionality in pandas

Problem Description

I have a dataset with date, industry, wage and price. I want to compute the percentage increase in wages and prices by industry. I can do something like:

data.set_index('date').groupby('industry')[['wage','price']].pct_change()

But this will return me a pandas dataframe with date as an index and wage and price as columns, losing the industry column.

Feature Description

It would be nice to add an option to pct_change(), when applied to a groupby object, to also return the groups, maybe as an index as the agg function does.

Alternative Solutions

My current approach is:
pd.concat([data.set_index('date')[['industry']], data.set_index('date').groupby(['industry'])[['wage','price']].pct_change()],axis=1)

Not a big deal but seems a common operation. Having a grouped dataframe and wanting to compute percentage changes for several columns.

Additional Context

No response

[rhshadrach]

Does

result = data.set_index('date').groupby('industry')[['wage','price']].pct_change()
result['industry'] = data['industry']

work? Expanding the applicability of as_index=False/True to transformations is part of #49543

[jesse-sealand]

@marcdelabarrera does @rhshadrach solution work for you?

[marcdelabarrera]

Yes, it would. A two-line option but it works. Thx!

[rhshadrach]

Using .assign(industry=data['industry']) would make it one line if you so prefer! Thanks for the response, closing.