![](/img/trans.png)
[英]Compare two dataframes, and then add new column to one of the data frames based on the other
[英]compare two data frames and add new column to dataframe based on mask values
我正在根據那里的ID比較兩個數據幀,然后使用以下代碼合並它們:
df = df1.merge(df2, on=id, suffixes=('_x','_y'))
df1
name age id salary
0 Smith 30 2 2000
1 Ron 24 3 30000
2 Mike 35 4 40000
3 Jack 21 5 5000
4 Roshan 20 6 60000
5 Steve 45 8 8000
6 Peter 28 1 1000
df2
name age salary id
0 Peter 32 10000 1
1 Smith 30 1500 2
2 Ron 24 7000 3
3 Mike 35 20000 4
4 Jack 21 5000 5
5 Cathy 20 9000 6
6 Steve 45 56000 8
o / p
name_x age_x id salary_x name_y age_y salary_y
0 Smith 30 2 2000 Smith 30 1500
1 Ron 24 3 30000 Ron 24 7000
2 Mike 35 4 40000 Mike 35 20000
3 Jack 21 5 5000 Jack 21 5000
4 Roshan 20 6 60000 Cathy 20 9000
5 Steve 45 8 8000 Steve 45 56000
6 Peter 28 1 1000 Peter 32 10000
現在基於輸出,我正在比較_x和_y列的值並將其放入mask中:
mask = df[cols + '_x'].values == df[cols + '_y'].values
print(mask)
面罩o / p
[[ True True False]
[ True True False]
[ True True False]
[ True True True]
[ True False False]
[ True True False]
[False True False]]
基於此掩碼值,我想提出以下條件:如果在mask [1]中出現false,它應該給我累積的“無匹配”結果,可以將其附加到輸出結果中,例如:
name_x age_x id salary_x name_y age_y salary_y new_column
0 Smith 30 2 2000 Smith 30 1500 No Match
1 Ron 24 3 30000 Ron 24 7000 No Match
2 Mike 35 4 40000 Mike 35 20000 No Match
3 Jack 21 5 5000 Jack 21 5000 MAtch
4 Roshan 20 6 60000 Cathy 20 9000 No Match
5 Steve 45 8 8000 Steve 45 56000 No Match
6 Peter 28 1 1000 Peter 32 10000 No Match
matches = ['Match' if x else 'No Match' for x in np.all(mask, axis = -1)]
將為您提供'Match'
和'No Match'
值的數組,您可以通過以下方式將其添加到數據框中:
df['newColumnName'] = matches
將numpy.where
與numpy.all
使用以進行快速矢量化解決方案:
mask = df[cols + '_x'].values == df[cols + '_y'].values
df['new_column'] = np.where(np.all(mask, axis=1) , 'Match','No Match')
print (df)
name_x age_x id salary_x name_y age_y salary_y new_column
0 Smith 30 2 2000 Smith 30 1500 No Match
1 Ron 24 3 30000 Ron 24 7000 No Match
2 Mike 35 4 40000 Mike 35 20000 No Match
3 Jack 21 5 5000 Jack 21 5000 Match
4 Roshan 20 6 60000 Cathy 20 9000 No Match
5 Steve 45 8 8000 Steve 45 56000 No Match
6 Peter 28 1 1000 Peter 32 10000 No Match
感謝您的評論@markuscosinus,如果需要通過索引索引將掩碼的第二個'column'
進行比較-此處通過mask[:, 1]
:
df['new_column'] = np.where(mask[:, 1] , 'Match','No Match')
將掩碼轉換為numpy數組或數據框,或者應該已經采用以下格式:
mask = pd.DataFrame([[ True, True, False],
[ True, True, False],
[ True, True, False],
[ True, True, True],
[ True, False, False],
[ True, True, False],
[False, True, False]])
然后下面的代碼為您提供所需的列:
mask.apply(sum, axis=1).apply(lambda x: 'Match' if x==3 else 'No Match')
您可以將此列添加到df
。
希望能幫助到你 ... :)
聲明:本站的技術帖子網頁,遵循CC BY-SA 4.0協議,如果您需要轉載,請注明本站網址或者原文地址。任何問題請咨詢:yoyou2525@163.com.