Question

I have a dataframe that holds data at a particular level of aggregation – let’s call it regional.

I also have a dict that explains how these regions are formed. Something like this:

map = {'Alabama': 'region_1', 'Arizona': 'region_1', 'Arkansas': 'region_2' ... }

And a set of weights for each state within its region, stored as a series:

Alabama    .25
Arizona    .75
Arkansas   .33
....

Is there an efficient way to apply this disaggregation map to get a new dataframe at a State level?

Aggregation is easy:

df_regional = df_states.groupby(map).sum()

But how can I do disaggregation?

Answer 1

Assuming two dataframes, df_states and df_regional, with the following
structure:

In [36]: df_states
Out[36]: 
          Weight    Region
Alabama     0.25  region_1
Arizona     0.75  region_1
Arkansas    0.33  region_2

In [37]: df_regional
Out[37]: 
          Value
region_1    100
region_2     80

Does pandas.merge arrange the data in a way that seems useful?

In [39]: df = pandas.merge(df_states, df_regional, left_on='Region', right_index=True)

In [40]: df
Out[40]: 
          Weight    Region  Value
Alabama     0.25  region_1    100
Arizona     0.75  region_1    100
Arkansas    0.33  region_2     80

In [41]: df.Weight * df.Value
Out[41]: 
Alabama     25.0
Arizona     75.0
Arkansas    26.4