logo标签列表

向 Pandas 数据框添加一个按排名排序的列

pythonpandasdataframe
4

我知道这个问题可能看起来很琐碎，但我找不到任何解决方案。我有一个非常大的Pandas数据框df，看起来像这样：

                                            conference     IF2013  AR2013
0                                            HOTMOBILE  16.333333   31.50
1                                                 FOGA  13.772727   60.00
2                                              IEA/AIE  10.433735   28.20
3    IEEE Real-Time and Embedded Technology and App...  10.250000   29.00
4                  Symposium on Computational Geometry   9.880342   35.00
5                                                 WISA   9.693878   43.60
6                                                 ICMT   8.750000   22.00
7                                              Haskell   8.703704   39.00

我想在最后添加一列，用1、2、3、4来排序。因此，它看起来像这样：
                                               conference     IF2013    AR2013  Ranking 
    0                                            HOTMOBILE  16.333333   31.50   1  
    1                                                 FOGA  13.772727   60.00   2
    2                                              IEA/AIE  10.433735   28.20   3
    3    IEEE Real-Time and Embedded Technology and App...  10.250000   29.00   4

我似乎无法找到方法来添加一个填充的额外列，该列只放置一系列连续的数字。

3个回答
6
您可以尝试：
df['rank'] = df.index + 1

print df
#                                          conference     IF2013  AR2013  rank
#0                                          HOTMOBILE  16.333333    31.5     1
#1                                               FOGA  13.772727    60.0     2
#2                                            IEA/AIE  10.433735    28.2     3
#3  IEEE Real-Time and Embedded Technology and App...  10.250000    29.0     4
#4                Symposium on Computational Geometry   9.880342    35.0     5
#5                                               WISA   9.693878    43.6     6
#6                                               ICMT   8.750000    22.0     7
#7                                            Haskell   8.703704    39.0     8

或者使用参数ascending=False的rank函数：

df['rank'] = df['conference'].rank(ascending=False)
print df
#                                          conference     IF2013  AR2013  rank
#0                                          HOTMOBILE  16.333333    31.5     1
#1                                               FOGA  13.772727    60.0     2
#2                                            IEA/AIE  10.433735    28.2     3
#3  IEEE Real-Time and Embedded Technology and App...  10.250000    29.0     4
#4                Symposium on Computational Geometry   9.880342    35.0     5
#5                                               WISA   9.693878    43.6     6
#6                                               ICMT   8.750000    22.0     7
#7                                            Haskell   8.703704    39.0     8
2
我猜您正在寻找rank函数：
df['rank'] = df['IF2013'].rank()

这样，您的结果将与索引无关。
1
你可以添加一个带有 range 的列：
df['Ranking'] = range(1, len(df) + 1)

例子：

import pandas as pd
from io import StringIO

data = """
                                        conference     IF2013  AR2013
                                        HOTMOBILE  16.333333   31.50
                                             FOGA  13.772727   60.00
                                          IEA/AIE  10.433735   28.20
IEEE Real-Time and Embedded Technology and App...  10.250000   29.00
              Symposium on Computational Geometry   9.880342   35.00
                                             WISA   9.693878   43.60
                                             ICMT   8.750000   22.00
                                          Haskell   8.703704   39.00

"""

df = pd.read_csv(StringIO(data), sep=' \s+')

df['Ranking'] = range(1, len(df) + 1)

In [183]: df
Out[183]:
                                          conference     IF2013  AR2013        Ranking  
0                                          HOTMOBILE  16.333333    31.5            1
1                                               FOGA  13.772727    60.0            2
2                                            IEA/AIE  10.433735    28.2            3
3  IEEE Real-Time and Embedded Technology and App...  10.250000    29.0            4
4                Symposium on Computational Geometry   9.880342    35.0            5
5                                               WISA   9.693878    43.6            6
6                                               ICMT   8.750000    22.0            7
7                                            Haskell   8.703704    39.0            8

编辑

基准测试：

In [202]: %timeit df['rank'] = range(1, len(df) + 1)
10000 loops, best of 3: 127 us per loop

In [203]: %timeit df['rank'] = df.AR2013.rank(ascending=False)
1000 loops, best of 3: 248 us per loop
网页内容由stack overflow 提供, 点击上面的
可以查看英文原文，
原文链接