比較兩個數據幀並檢索公共行元素

Question

我需要比較兩個數據集：

DF1

       Subj             1           2           3
0   Biotech   Cell culture     Bioinfo  Immunology
1   Zoology   Cell culture  Immunology         NaN
2      Math   Trigonometry     Algebra         NaN
3  Microbio        Biotech         NaN         NaN
4   Physics         Optics         NaN         NaN

DF2

       Subj             1           2           
0   Biotech       Bioinfo  Immunology         
1   Zoology    Immunology      Botany                  
2  Microbio         NaN           NaN         
3   Physics        Optics  Quantumphy
4      Math  Trigonometry         NaN         

我如何想要我的結果 dataframe：

       Subj             1           2          
0   Biotech       Bioinfo  Immunology         
1   Zoology    Immunology         NaN         
2      Math  Trigonometry         NaN         
3   Physics        Optics         NaN         

由於數據集很大，我無法逐行檢查。 兩個數據集的列數不同，但行數相同。 由於行元素的順序也不同，我不能簡單地使用 merge()。 我嘗試比較 function，但它要么刪除所有公共元素，要么 forms 和包含這兩者的 dataframe。 我似乎不能只挑出共同的元素。

Answer 1

您可以匹配列，然后在合並數據幀時將主題列設置為索引：

match=df2.columns.intersection(df1.columns).tolist()
df2.merge(df1,on=match, how='left').reindex(df2.columns,axis=1).set_index('Subj').dropna(how='all')

返回：

                    1           2
Subj                             
Biotech       Bioinfo  Immunology
Zoology    Immunology         NaN
Math     Trigonometry         NaN
Physics        Optics         NaN

Answer 2

這是一種方法

理解：列數不同並且兩個 DF 中的值不在同一列下

# Stack both the DFs, after setting Subj as index
# this results in changing a wide format to long format
# concat the two DF to forma new DF

df3=pd.concat([df.set_index('Subj').stack().reset_index().rename(columns={0:'val'}),
          df2.set_index('Subj').stack().reset_index().rename(columns={0:'val'})],
          ).reset_index()


# to find the same topic under a subject if it exists in two DFs
# the join will have duplicate rows

# so find the duplicated rows for Subj and Topic (val column)
# group the duplicated rows and aggregate to a comma separated values
# finally split on comma to create new columns

out=(df3[df3.duplicated(subset=['Subj','val'])]
 .groupby('Subj')['val']
 .agg(','.join)
 .str
 .split(',',expand=True).reset_index())
out

    Subj        0             1
0   Biotech     Bioinfo       Immunology
1   Math        Trigonometry  None
2   Physics     Optics        None
3   Zoology     Immunology    None