Dataframe 到字典分组和键/值是其他列

Question

Back my on bs, friends. I have a dataframe like so:

+-------+--------------------------+------+-------+------+
| index |                specialty | code | count | rank |
+-------+--------------------------+------+-------+------+
| 19    | Colon and Rectal Surgery | 1557 | 36    | 5.0  |
+-------+--------------------------+------+-------+------+
| 22    | Surgical Oncology        | 1557 | 22    | 14.0 |
+-------+--------------------------+------+-------+------+
| 147   | Hematology               | 2057 | 383   | 13.0 |
+-------+--------------------------+------+-------+------+
| 753   | Oncology                 | 1578 | 74    | 15.0 |
+-------+--------------------------+------+-------+------+
| 1089  | Dental General Practice  | 1257 | 6     | 2.5  |
+-------+--------------------------+------+-------+------+

There are multiple entries per specialty -- ie I have the count and rank of codes for Specialty X up to Rank 25.

I'm trying to use a lamba function to group by specialty but I can't figure out how to add the columns as the keys/values and create a list of dict rather than just a giant dict.

d = (df2.groupby('specialty').apply(lambda x: dict(zip(x['code'], x['Rank']))).to_dict())

print(d)

{'Acute Care Hospital': {
    1562: 8.0, 
    1554: 11.0, 
    6095: 8.0, 
    119114: 1.0, 
    119117: 5.5, 
    284051: 4.0, 
    562577: 11.0, 
    582646: 8.0, 
    1631305: 2.0, 
    1641114: 5.5, 
    1751592: 3.0, 
    1873207: 11.0
}

How do I get preserve the columns as the keys like so and it be a list per specialty:

[
    {'specialty': Acute Care Hospital', 
    [
        {'code': 1562, 'rank': 8.0, 
        'code': 1554, 'rank' :11.0, 
        'code': 6095, 'rank': 8.0, 
        'code': 119114, 'rank' 1.0, 
        'code': 119117, 'rank': 5.5, 
        'code': 284051, 'rank': 4.0, 
        'code': 562577, 'rank': 11.0, 
        'code': 582646, 'rank' 8.0, 
        'code': 1631305, 'rank': 2.0, 
        'code': 1641114, 'rank': 5.5, 
        'code': 1751592, 'rank': 3.0, 
        'code': 1873207, 'rank': 11.0}
    ]
    }
]

Answer 1

The outcome that you posted will not work as it contains duplicate keys. The below solutions might be what you are after as it makes the code rank pairs callable from the dictionary.

This one creates another dictionary level underneath the main level speciality, where the code and rank pairs are on the same index in the arrays:

df.groupby('specialty').apply(lambda x: {'code':x['code'].values,'Rank':x['Rank'].values}).to_dict()

Or the next one simply gets the result from the groupby and places that under the key code_rank_pair:

df.groupby('specialty').apply(lambda x: {'code_rank_pair':x.loc[:,['code','Rank']]}).to_dict()