python：将组大小连接到数据框中的成员行

Question

(Python 2.7) I wish to create a column in a python dataframe with the size of the group to which member rows belong (indexed by row ID number). Groups are based on rows with identical values in two columns, date and amount. I've attempted to use groubpy and size - which is suggested for similar problems - but I can't get the resulting size values back to the source dataframe due to indexing problems. Should I use a dictionary to read all unique value pairings instead, and what would that look like? Or should I learn how to merge the groupby object to the original dataframe with a join operation. Note: this is large dataset.

Sample data:

                    date    amount  address
    ID          
    176820  1/4/2008 0:00   400     13496 ST LOUIS
    176821  1/4/2008 0:00   500     13475 NEWBERN
    176822  1/4/2008 0:00   2000    8011 DAYTON
    176823  1/4/2008 0:00   4000    13406 LONGVIEW
    176824  1/4/2008 0:00   7000    19174 ARCHDALE

Here's what I thought might work:

    df['group_size'] = df.groupby(['date','amount']).size()

But I received this: TypeError: incompatible index of inserted column with frame index

UPDATE: elyase's solution works for the original sample data I posted. My source dataframe actually has 13 columns, not 3, but elyase's solution doesn't work when even one additional column is added to the sample frame.

                     date  amount         address    tract
    ID                                                    
    176820  1/4/2008 0:00     400  13496 ST LOUIS   510200
    176821  1/4/2008 0:00     500   13475 NEWBERN   510400
    176822  1/4/2008 0:00    2000     8011 DAYTON   526200
    176823  1/4/2008 0:00    4000  13406 LONGVIEW   504200
    176824  1/4/2008 0:00    7000  19174 ARCHDALE   540200

I get the error: Wrong number of items passed 1, indices imply 2

Answer 1

你有没有尝试过：

df.groupby(['date','amount']).transform('count')

Answer 2

To get the group count, I needed to count over any OTHER variable in my group. The only issue here is that where the amount column is null, size returns the tract value, but this is easily dealt with.

    df['group_size'] = df.groupby(['date','amount'])['tract'].transform('count')