Question

我的数据看起来像

 Time 	          Pressure	Normal/Abnormal
11/30/2011 22:50	74.3	  0
11/30/2011 23:00	74.8	  1
11/30/2011 23:10	77.7	  1
11/30/2011 23:30	74.8	  0
11/30/2011 13:00	80.9	  0

Desired Output:

Time 	           Normal	Time 	           Abnormal
11/30/2011 22:50	74.3	11/30/2011 23:00	74.8
11/30/2011 23:30	74.8	11/30/2011 23:10	77.7
11/30/2011 13:00	80.9

我想像在“期望的输出”中提到的那样转置行。我知道我需要使用类似于melt and cast（在R中使用）的东西，但是不确定如何使用它们。

Answer 1

使用上面的数据

import pandas as pd
from io import StringIO
import itertools

text = u'Time \t          Pressure\tNormal/Abnormal\n11/30/2011 22:50\t74.3\t  0\n11/30/2011 23:00\t74.8\t  1\n11/30/2011 23:10\t77.7\t  1\n11/30/2011 23:30\t74.8\t  0\n11/30/2011 13:00\t80.9\t  0'

df = pd.read_table(StringIO(text))
normal = df.loc[df['Normal/Abnormal'] == 0].as_matrix()
abnormal = df.loc[df['Normal/Abnormal'] == 1].as_matrix()

columns = ["Time", "Normal", "Time", "Abnormal"]
out = []

for nr, ar in itertools.izip_longest(normal, abnormal, fillvalue=['', '']):
    # Concat rows horizontally (i.e. hstack)
    r = list(nr[:2]) + list(ar[:2])
    out.append(r)

df2 = pd.DataFrame(out, columns=columns)

print df2.to_string(index=False)

''' Output
Time  Normal              Time Abnormal
11/30/2011 22:50    74.3  11/30/2011 23:00     74.8
11/30/2011 23:30    74.8  11/30/2011 23:10     77.7
11/30/2011 13:00    80.9
'''

Answer 2

构造两个数据帧，其中1个表示正常，1个表示异常，然后concat并编辑列名

out = pd.concat([
  df[df['Normal/Abnormal'] == k].iloc[:, [0,1]].reset_index(drop=True)
  for k in [0, 1]], axis=1
)
out.columns = ['Time', 'Normal', 'Time', 'Abnormal']
out
               Time  Normal              Time  Abnormal
0  11/30/2011 22:50    74.3  11/30/2011 23:00      74.8
1  11/30/2011 23:30    74.8  11/30/2011 23:10      77.7
2  11/30/2011 13:00    80.9               NaN       NaN

用实际值替换Python分数（转置）

2 个答案: