使用 fuzzywuzzy 完成任务的软件包：

from fuzzywuzzy import process

THRESHOLD = 80

best_match = df2['Full name'].apply(lambda x: process.extractOne(x, df1['Full name'],
                                                                 score_cutoff=THRESHOLD))

df2['match'] = best_match.apply(pd.Series)[0]

输出：

>>> df2
            Full name            match
0    daniel lastname1    dan lastname1
1  william lastname 2  bill lastname 2
2    robert lastname3    bob lastname3
3  bradley lastname 4  brad lastname 4
4     Jane lastname 5              NaN

>>> best_match
0      (dan lastname1, 90, 0)
1    (bill lastname 2, 85, 2)
2      (bob lastname3, 83, 3)
3    (brad lastname 4, 91, 4)
4                        None
Name: Full name, dtype: object

如何安装

pip install fuzzywuzzy python-levenshtein
conda install -c conda-forge fuzzywuzzy python-levenshtein

这个 python-levenshtein 软件包不是强制性的，但它在字符串匹配中提供了4-10倍的加速。