python - 熊猫：绕过数字索引

Question

有时我正在处理具有DataFrame数字索引的 s，但我想绕过它以根据它们的顺序引用行，

In [49]: df = pandas.DataFrame(np.random.randn(3, 5))

In [50]: df
Out[50]: 
          0         1         2         3         4
0 -2.426211  0.670384  0.545880 -1.435168  0.675598
1  0.507128  0.478832 -0.159536 -0.696284 -1.112171
2  0.938019 -1.673491 -0.567462  0.381804 -1.280602

In [51]: df[1:2]
Out[51]: 
          0         1         2         3         4
1  0.507128  0.478832 -0.159536 -0.696284 -1.112171

In [52]: df2 = df.ix[1:2]

In [53]: df2
Out[53]: 
          0         1         2         3         4
1  0.507128  0.478832 -0.159536 -0.696284 -1.112171
2  0.938019 -1.673491 -0.567462  0.381804 -1.280602

In [55]: df2.ix[0]
---------------------------------------------------------------------------
KeyError                                  Traceback (most recent call last)
...

In [56]: df2.rename(index={1: "one", 2:"two"})
Out[56]: 
            0         1         2         3         4
one  0.507128  0.478832 -0.159536 -0.696284 -1.112171
two  0.938019 -1.673491 -0.567462  0.381804 -1.280602

In [57]: df3 = df2.rename(index={1: "one", 2:"two"})

In [58]: df3.ix[0]
Out[58]: 
0    0.507128
1    0.478832
2   -0.159536
3   -0.696284
4   -1.112171
Name: one

如何绕过 pandas 的索引来访问底层矩阵索引？

Answer 1

您可以使用 DataFrame.irow：

In [18]: df2
Out[18]: 
          0         1         2
1  2.279885 -0.414938 -2.230296
2 -0.237980 -0.219556  1.231576

In [19]: df2.irow(0)
Out[19]: 
0    2.279885
1   -0.414938
2   -2.230296
Name: 1

In [20]: df2.irow([0, 1])
Out[20]: 
          0         1         2
1  2.279885 -0.414938 -2.230296
2 -0.237980 -0.219556  1.231576