Pandas Compare(), как перебирать строки

Вы можете использовать этот подход:

import pandas as pd

df1 = pd.DataFrame({"userid": [122, 123, 124], "amount": [1, 2, 3]})

df2 = pd.DataFrame({"userid": [121, 123, 124], "amount": [1, 2, 4]})

updated_records = df1.compare(df2)

main_cols = set([col[0] for col in updated_records.columns])

for col in main_cols:
    updated_records[f"{col}_comp"] = updated_records.apply(
        lambda row: f'{row[(col, "self")]} -> {row[(col, "other")]}', axis=1
    )

for i, row in updated_records.iterrows():
    print(f"Fields updated on row {i}:")
    for col in main_cols:
        updated = row[f"{col}_comp"].iloc[0]
        if "nan" not in updated:
            print(f"{col}: {updated}")
    print()

Fields updated on row 0:
userid: 122.0 -> 121.0

Fields updated on row 2:
amount: 3.0 -> 4.0

Пример кода

Было неясно, какой именно стиль вывода вам нужен, поэтому я привел его в качестве нового примера.

import pandas as pd
data = {
    ('userid', 'self'): [122, None],
    ('userid', 'other'): [121, None],
    ('amount', 'self'): [None, 3.0],
    ('amount', 'other'): [2, 4.0]
}

df = pd.DataFrame(data, index=[0, 2])

дф:

Код

for idx, row in df.iterrows():
    print('For item {} those fields where updated:'.format(idx))
    colums_lv0 = row[row.notna()].index.get_level_values(0).drop_duplicates()
    for col in colums_lv0:        
        print('{} : {} -> {}'.format(col, row[(col, 'self')], row[(col, 'other')]))
    print()

Распечатать:

For item 0 those fields where updated:
userid : 122.0 -> 121.0
amount : nan -> 2.0

For item 2 those fields where updated:
amount : 3.0 -> 4.0