熊猫MultiIndex数据框排序

Question

I am looking for a way to sort a multiindex data frame based on one of the indices. Here's an example ...

I have ...

mi = pd.MultiIndex.from_tuples([('baz', 'y'), ('foo', 'y'), ('bar', 'y'), ('baz', 'z'), ('foo', 'z'), ('bar', 'z')],names=['first','second'])
df = pd.DataFrame(np.random.randn(6),index=mi)
print df

which produces

first second          
baz   y       1.902902
foo   y      -0.128341
bar   y       0.481100
baz   z      -2.185144
foo   z       1.015320
bar   z      -1.624616

If I do ...

df2 = df.sort()

I get :

print df2

first second          
bar   y       0.481100
      z      -1.624616
baz   y       1.902902
      z      -2.185144
foo   y      -0.128341
      z       1.015320

How can I sort df such that I get :

first second          
baz   y       1.902902
      z      -2.185144
foo   y      -0.128341
      z       1.015320
bar   y       0.481100
      z      -1.624616

Answer 1

I'm not sure how bar is sorted after foo, but I think this is what you want:

df.sortlevel(0)

                     0
first second          
bar   y      -0.114659
      z       0.453896
baz   y      -0.327006
      z       0.685782
foo   y       0.444256
      z      -0.167179