[英]Pandas Compare between Two DataFrames, flag what matches
我必须要数据帧df
和df1
df
低于
Facility Category ID Part Text
Centennial History 11111 A Drain
Centennial History 11111 B Read
Centennial History 11111 C EKG
Centennial History 11111 D Assistant
Centennial History 11111 E Primary
df1
在下面(仅包含一个小样本问题,实际上是50,000行)
Facility Category ID Part Text
Centennial History 11111 D Assistant
基本上,我想比较数据框之间的行,如果行在两个数据框之间匹配,则在第一个数据框df
创建另一个列,列标题为['MatchingFlag']
我的最终结果数据框如下所示,因为我担心那些不匹配的数据框。
Facility Category ID Part Text MatchingFlag
Centennial History 11111 A Drain No
Centennial History 11111 B Read No
Centennial History 11111 C EKG No
Centennial History 11111 D Assistant Yes
Centennial History 11111 E Primary No
有什么帮助吗? 我试过合并df = pd.merge(df1, df, how='left', on=['Facility', 'Category', 'ID', 'Part', 'Text'])
然后根据空白或NaN值创建一个标志,但这并没有达到我的期望。
在要匹配的列上设置索引,然后使用该索引来排序匹配的行可能是有意义的
columns = ['Facility', 'Category', 'ID', 'Part', 'Text']
# It's always a good idea to sort after creating a MultiIndex like this
df = df.set_index(columns).sortlevel()
df1 = df1.set_index(columns).sortlevel()
# You don't have to use Yes here, anything will do
# The boolean True might be more appropriate
df['MatchingFlag'] = "Yes"
df1['MatchingFlag'] = "Yes"
# Add them together, matching rows will have the value "YesYes"
# Non-matches will be nan
result = df + df1
# If you'd rather not have NaN's
result.loc[:,'MatchingFlag'] = result.loc[:,'MatchingFlag'].replace('YesYes','Yes')
result.loc[:,'MatchingFlag'] = result['MatchingFlag'].fillna('No')
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.