Question

我正在寻找一种更具编程性的方法来创建多个新列，作为Pandas DataFrame中现有列的函数。

我有14列Level_2 - Level_15。我想迭代地创建14个新列，它们将列2-15，然后是3-15，然后是4-15，依此类推。

现在我的代码看起来像这样

cols['2_sum'] = cols.Level_2 + cols.Level_3 + cols.Level_4 + cols.Level_5 + cols.Level_6 + cols.Level_7 + cols.Level_8 + cols.Level_9 + cols.Level_10 + cols.Level_11 + cols.Level_12 + cols.Level_13 + cols.Level_14 + cols.Level_15
cols['3_sum'] = cols.Level_3 + cols.Level_4 + cols.Level_5 + cols.Level_6 + cols.Level_7 + cols.Level_8 + cols.Level_9 + cols.Level_10 + cols.Level_11 + cols.Level_12 + cols.Level_13 + cols.Level_14 + cols.Level_15
cols['4_sum'] = cols.Level_4 + cols.Level_5 + cols.Level_6 + cols.Level_7 + cols.Level_8 + cols.Level_9 + cols.Level_10 + cols.Level_11 + cols.Level_12 + cols.Level_13 + cols.Level_14 + cols.Level_15

是否有更多的熊猫或pythonic方式来做到这一点？

谢谢！

Answer 1

以下是一个例子：

示例数据：

In [147]: df = pd.DataFrame(np.random.rand(3, 15),
     ...:                   columns=['ID'] + ['Level_{}'.format(x) for x in range(2, 16)])
     ...:

In [148]: df
Out[148]:
         ID   Level_2   Level_3   Level_4   Level_5   Level_6   Level_7   Level_8   Level_9  Level_10  Level_11  \
0  0.851407  0.957810  0.204217  0.848265  0.168324  0.010265  0.191499  0.787552  0.648678  0.424462  0.038888
1  0.354270  0.442843  0.631624  0.081120  0.357300  0.211621  0.177321  0.316312  0.836935  0.445603  0.267165
2  0.998240  0.341875  0.590768  0.475935  0.071915  0.720590  0.041327  0.926167  0.671880  0.516845  0.450720

   Level_12  Level_13  Level_14  Level_15
0  0.465109  0.508491  0.282262  0.848373
1  0.205415  0.399493  0.537186  0.774417
2  0.131734  0.554596  0.253658  0.104193

解决方案：

In [149]: for n in range(15, 1, -1):
     ...:     df['{}_sum'.format(15-n+2)] = df.filter(regex=r'Level_\d+').iloc[:, :n].sum(1)
     ...:

结果：

In [150]: df
Out[150]:
         ID   Level_2   Level_3   Level_4   Level_5   Level_6   Level_7   Level_8   Level_9  Level_10    ...     \
0  0.851407  0.957810  0.204217  0.848265  0.168324  0.010265  0.191499  0.787552  0.648678  0.424462    ...
1  0.354270  0.442843  0.631624  0.081120  0.357300  0.211621  0.177321  0.316312  0.836935  0.445603    ...
2  0.998240  0.341875  0.590768  0.475935  0.071915  0.720590  0.041327  0.926167  0.671880  0.516845    ...

      6_sum     7_sum     8_sum     9_sum    10_sum    11_sum    12_sum    13_sum    14_sum    15_sum
0  4.745067  4.279958  4.241070  3.816608  3.167931  2.380379  2.188880  2.178615  2.010292  1.162027
1  3.973259  3.767844  3.500679  3.055076  2.218140  1.901828  1.724508  1.512887  1.155587  1.074468
2  4.939755  4.808021  4.357301  3.840456  3.168576  2.242409  2.201082  1.480492  1.408577  0.932643

[3 rows x 29 columns]

Answer 2

您可以创建列列表

cols= list(cols)
cols['2_sum'] = cols[cols].sum(axis = 1)
cols['3_sum'] = cols['2_sum'] - cols['Level_2']
cols['4_sum'] = cols['3_sum'] - cols['Level_3']

Answer 3

希望它可以帮到你

ColsListName = ['3_sum' ... ,'14_sum']
ColsListLevel = ['Level_2','Level_3' ... ,'Level_15']
sumCols = cols.Level_2 + cols.Level_3 + cols.Level_4 + cols.Level_5 + cols.Level_6 + cols.Level_7 + cols.Level_8 + cols.Level_9 + cols.Level_10 + cols.Level_11 + cols.Level_12 + cols.Level_13 + cols.Level_14 + cols.Level_15
cols['2_sum'] = sumCols 
for i in range(len(ColsListLevel)) :
   cols[ColsListName [i]] = sumCols - cols.ColsListLevel [i]

Answer 4

import pandas as pd
import numpy as np
np.random.seed(1)
cols = pd.DataFrame(np.random.rand(2, 14), 
                    columns=['Level_'+str(i) for i in range(2, 16)])

现在数据框看起来像：

    Level_2     Level_3     Level_4     Level_5     Level_6     Level_7     Level_8     Level_9     Level_10    Level_11    Level_12    Level_13    Level_14    Level_15    2_sum
0   0.199666    0.285152    0.598139    0.602477    0.004284    0.874587    0.263949    0.527301    0.306443    0.282778    0.181330    0.280506    0.456637    0.998124    5.861371
1   0.279320    0.508074    0.435350    0.816866    0.691988    0.179261    0.134478    0.949185    0.867022    0.410112    0.139481    0.537539    0.042163    0.366138    6.356977

然后：

for i in range(2, 15):
    cols[str(i)+'_sum'] = cols.loc[:, 'Level_'+str(i):'Level_15'].sum(axis=1)
cols

我认为这就是你想要的。

作为pandas中现有列的函数创建新列的动态方法

4 个答案: