How to reorder the columns of a MultiIndex columns pandas DataFrame?-CodePudding

After having read the pandas reindex docs, it is still not clear to me how to re-arrange the order of the columns in a MultiIndex DataFrame and re-assign it to the original DataFrame. For example, say that I have the following DataFrame

df = 
kind                            A               B
names          u1      u2      u3      y1      y2
Time
0.0        0.5083  0.1007  0.8001  0.7373  0.1387
0.1        0.6748  0.0354  0.0076  0.8421  0.2670
0.2        0.1753  0.1013  0.5231  0.8060  0.0040

and I want to re-order some columns such that at the end I get the following

df = 
kind                            A               B
names          u3      u2      u1      y1      y2
Time
0.0        0.8001  0.1007  0.5083  0.7373  0.1387
0.1        0.0076  0.0354  0.6748  0.8421  0.2670
0.2        0.5231  0.1013  0.1753  0.8060  0.0040

If I do df.reindex(level="names", columns=["u3","u1"]), then I got

kind                    A      
names          u3      u1      
Time
0.0        0.8001  0.5083  
0.1        0.0076  0.6748  
0.2        0.5231  0.1753

which is encouraging, but then I cannot figure out how to re-assign it to the original df. I tried df.loc[:,"A"] = df.reindex(level="names", columns=["u3","u1"]) but what I get is the following

kind                          A               B
names          u1    u2      u3      y1      y2
Time
0.0        0.5083   NaN  0.8001  0.7373  0.1387  
0.1        0.6748   NaN  0.0076  0.8421  0.2670
0.2        0.1753   NaN  0.5231  0.8060  0.0040

CodePudding user response：

Here is solution for recreate MultiIndex with expected ordering:

print (df)
kind        C       A                       B
names      d1      u2      u3      u1      y2
0.0    0.5083  0.1007  0.8001  0.7373  0.1387
0.1    0.6748  0.0354  0.0076  0.8421  0.2670
0.2    0.1753  0.1013  0.5231  0.8060  0.0040

#original columns
cols = df.columns

#columns after change order
new = df.reindex(level="names", columns=["u3","u2", "u1"]).columns

#filter only tuples from new
d = dict(zip(cols[cols.isin(new)], new))

#recreate Multiindex with mapping and reindex all columns
mux = pd.MultiIndex.from_tuples([d.get(x, x) for x in cols],names=["kind","names"])
df = df.reindex(mux, axis=1)
print (df)
kind        C       A                       B
names      d1      u3      u2      u1      y2
0.0    0.5083  0.8001  0.1007  0.7373  0.1387
0.1    0.6748  0.0076  0.0354  0.8421  0.2670
0.2    0.1753  0.5231  0.1013  0.8060  0.0040

CodePudding user response：

df2 = df.reindex(level="names", columns=["u3", "u2", "u1", "y1", "y2"])