我有两个 pandas 数据框，如下所示:
df_out:
Prediction count_human count_bot %_bot_tweets
username
666STEVEROGERS 8 131 0.942446
ADELE_BROCK 0 126 1.000000
ADRIANAMFTTT 99 0 0.000000
AHMADRADJAB 0 108 1.000000
ALBERTA_HAYNESS 101 0 0.000000
ALTMANBELINDA 0 139 1.000000
ALVA_MC_GHEE 29 104 0.781955
ANGELITHSS 0 113 1.000000
ANN1EMCCONNELL 0 125 1.000000
ANWARJAMIL22 0 112 1.000000
AN_N_GASTON 0 107 1.000000
ARONHOLDEN8 89 31 0.258333
ARTHCLAUDIA 0 103 1.000000
ASSUNCAOWALLAS 0 108 1.000000
BECCYWILL 0 132 1.000000
BELOZEROVNIKIT 132 8 0.057143
BEN_SAR_GENT 24 84 0.777778
BERT_HENLEY 105 0 0.000000
BISHOLORINE 0 117 1.000000
BLACKERTHEBERR5 4 100 0.961538
BLACKTIVISTSUS 49 68 0.581197
BLACK_ELEVATION 32 74 0.698113
BOGDANOVAO2 0 127 1.000000
BREMENBOTE 70 39 0.357798
B_stever96 0 171 1.000000
CALIFRONIAREP 60 72 0.545455
C_dos_94 0 121 1.000000
Cassidygirly 0 153 1.000000
ChuckSpeaks_ 0 185 1.000000
Cyabooty 111 0 0.000000
DurkinSays 0 131 1.000000
LSU_studyabroad 117 0 0.000000
MisMonWEXP 131 0 0.000000
NextLevel_Mel 0 185 1.000000
PeterDuca 108 0 0.000000
ShellMarcel 0 97 1.000000
Sir_Fried_Alott 0 144 1.000000
XavierRivera_ 197 0 0.000000
ZacharyFlair 213 0 0.000000
brentvarney44 0 126 1.000000
cbars68 225 0 0.000000
chloeschultz11 0 106 1.000000
hoang_le_96 0 104 1.000000
kdougherty178 0 127 1.000000
lasallephilo 138 0 0.000000
lovely_cunt_ 0 137 1.000000
megliebsch 0 217 1.000000
msimps_15 138 0 0.000000
okweightlossdna 105 0 0.000000
tankthe_hank 231 0 0.000000
和knn_res:
following followers username Prediction is_bot
0 199 77 megliebsch 1 0
1 199 77 megliebsch 1 0
2 199 77 megliebsch 1 0
3 199 77 megliebsch 1 0
4 199 77 megliebsch 1 0
... ... ... ... ... ...
6643 67 57 ASSUNCAOWALLAS 1 1
6644 67 57 ASSUNCAOWALLAS 1 1
6645 67 57 ASSUNCAOWALLAS 1 1
6646 67 57 ASSUNCAOWALLAS 1 1
6647 67 57 ASSUNCAOWALLAS 1 1
我想做的是，对于df_out中的每个用户名，左连接到knn_res以获得以下 和 followers 值。
在 SQL 中，我可以通过以下方式执行此操作:
从 df_out a LEFT JOIN knn_res b ON a.username = b.username 中选择 a.*、b.following、b.followers
我已经尝试过:
test_df = df_out
test_df.set_index('username').join(knn_res.set_index('username'), on='username', how='left')
print(test_df)
结果:
File "C:\Python367-64\lib\site-packages\pandas\core\frame.py", line 4396, in set_index
raise KeyError("None of {} are in the columns".format(missing))
KeyError: "None of ['username'] are in the columns"
我做错了什么？我试图引用this documentation for the problem .
更新
我还尝试了内部联接，它产生了完全相同的结果:
File "C:\Python367-64\lib\site-packages\pandas\core\frame.py", line 4396, in set_index
raise KeyError("None of {} are in the columns".format(missing))
KeyError: "None of ['username'] are in the columns"
df_out 的创建方式为:
df_out = (knn_res.groupby(['username', 'Prediction']).is_bot.count().unstack(fill_value=0).
rename({0: 'count_human', 1: 'count_bot'}, axis= 1))
df_out['%_bot_tweets'] = df_out['count_bot'] / (df_out['count_bot'] + df_out['count_human'])
最佳答案
试试这个。默认 join 选项为 left，因此您无需指定它。两个数据帧都将 username 作为索引，并且 join 适用于索引，因此您也不需要指定 on 选项。最后，您只想连接 following 和 followers 列，因此在将 username 设置为索引后，只需对这两列进行切片即可连接。 (注意:当您想要将原始数据帧复制到 test_df 时，您应该使用 copy()，因为如果没有 copy()，两者指向同一个数据框对象)
test_df = df_out.copy()
test_df = test_df.join(knn_res.set_index('username')[['following', 'followers']])
print(test_df)
Out[93]:
count_human count_bot %_bot_tweets following followers
username
666STEVEROGERS 8 131 0.942446 NaN NaN
ADELE_BROCK 0 126 1.000000 NaN NaN
ADRIANAMFTTT 99 0 0.000000 NaN NaN
AHMADRADJAB 0 108 1.000000 NaN NaN
ALBERTA_HAYNESS 101 0 0.000000 NaN NaN
ALTMANBELINDA 0 139 1.000000 NaN NaN
ALVA_MC_GHEE 29 104 0.781955 NaN NaN
ANGELITHSS 0 113 1.000000 NaN NaN
ANN1EMCCONNELL 0 125 1.000000 NaN NaN
ANWARJAMIL22 0 112 1.000000 NaN NaN
AN_N_GASTON 0 107 1.000000 NaN NaN
ARONHOLDEN8 89 31 0.258333 NaN NaN
ARTHCLAUDIA 0 103 1.000000 NaN NaN
ASSUNCAOWALLAS 0 108 1.000000 67.0 57.0
ASSUNCAOWALLAS 0 108 1.000000 67.0 57.0
ASSUNCAOWALLAS 0 108 1.000000 67.0 57.0
ASSUNCAOWALLAS 0 108 1.000000 67.0 57.0
ASSUNCAOWALLAS 0 108 1.000000 67.0 57.0
BECCYWILL 0 132 1.000000 NaN NaN
BELOZEROVNIKIT 132 8 0.057143 NaN NaN
BEN_SAR_GENT 24 84 0.777778 NaN NaN
BERT_HENLEY 105 0 0.000000 NaN NaN
BISHOLORINE 0 117 1.000000 NaN NaN
BLACKERTHEBERR5 4 100 0.961538 NaN NaN
BLACKTIVISTSUS 49 68 0.581197 NaN NaN
BLACK_ELEVATION 32 74 0.698113 NaN NaN
BOGDANOVAO2 0 127 1.000000 NaN NaN
BREMENBOTE 70 39 0.357798 NaN NaN
B_stever96 0 171 1.000000 NaN NaN
CALIFRONIAREP 60 72 0.545455 NaN NaN
C_dos_94 0 121 1.000000 NaN NaN
Cassidygirly 0 153 1.000000 NaN NaN
ChuckSpeaks_ 0 185 1.000000 NaN NaN
Cyabooty 111 0 0.000000 NaN NaN
DurkinSays 0 131 1.000000 NaN NaN
LSU_studyabroad 117 0 0.000000 NaN NaN
MisMonWEXP 131 0 0.000000 NaN NaN
NextLevel_Mel 0 185 1.000000 NaN NaN
PeterDuca 108 0 0.000000 NaN NaN
ShellMarcel 0 97 1.000000 NaN NaN
Sir_Fried_Alott 0 144 1.000000 NaN NaN
XavierRivera_ 197 0 0.000000 NaN NaN
ZacharyFlair 213 0 0.000000 NaN NaN
brentvarney44 0 126 1.000000 NaN NaN
cbars68 225 0 0.000000 NaN NaN
chloeschultz11 0 106 1.000000 NaN NaN
hoang_le_96 0 104 1.000000 NaN NaN
kdougherty178 0 127 1.000000 NaN NaN
lasallephilo 138 0 0.000000 NaN NaN
lovely_cunt_ 0 137 1.000000 NaN NaN
megliebsch 0 217 1.000000 199.0 77.0
megliebsch 0 217 1.000000 199.0 77.0
megliebsch 0 217 1.000000 199.0 77.0
megliebsch 0 217 1.000000 199.0 77.0
megliebsch 0 217 1.000000 199.0 77.0
msimps_15 138 0 0.000000 NaN NaN
okweightlossdna 105 0 0.000000 NaN NaN
tankthe_hank 231 0 0.000000 NaN NaN
关于python - 连接两个 pandas 数据框失败，我们在Stack Overflow上找到一个类似的问题： https://stackoverflow.com/questions/59005016/