
        
        
            
                
                
                    index = [("A", "alice"), ("A", "bob"),("A", "cathy"),("B", "david"),("B", "eric")]
multi_index = pd.MultiIndex.from_tuples(index)
df = pd.DataFrame({"a":[2,3,4,5,6]}, index=multi_index)
df
                
            
                        a
A   alice   2
    bob     3
    cathy   4
B   david   5
    eric    6

To swap the ordering of the row levels:


        
        
            
                
                
                    df.reorder_levels([1,0])
                
            
                        a
alice   A   2
bob     A   3
cathy   A   4
david   B   5
eric    B   6

Here, the argument [1,0] means the following:

1st level (inner level in this case) becomes the outer level.
0th level (outer level) becomes the inner level.

Reordering levels of multi-index column

Consider the following DataFrame with multi-index columns:


        
        
            
                
                
                    index = [("A", "alice"), ("A", "bob"),("A", "cathy"), ("B", "david"),("B", "eric")]
multi_index = pd.MultiIndex.from_tuples(index)
df = pd.DataFrame([[2,3,4,5,6]], columns=multi_index)
df
                
            
                A                     B
    alice   bob   cathy   david   eric
0   2       3     4       5       6

To swap the ordering of the column levels, set axis=1 like so:


        
        
            
                
                
                    df.reorder_levels([1,0], axis=1)
                
            
                alice   bob   cathy   david   eric
    A       A     A       B       B
0   2       3     4       5       6