Question

我有一个csv，格式如下

       Time     Marker
0       2104    21
1       2109    20
2       2485    21
3       2491    20
4       2867    22
5       2997    2
6       3248    23

我想计算Marker == 20之间21,22和23s的发生率。唯一有效的标记在20个代码之间预订，因此前21个无效。多个有效标记可以出现在已预订的20对中，因此我需要在一对20s之间出现21,22和23s的计数。

因此，在上面的示例中，只有索引2可以是有效代码，因为它介于两个20之间。

我有一个满足Marker == 20条件的索引列表

Indexrange = df.index[df['Marker'] == 20].tolist()
[1,
 3,
 10,
 19,
 22,
 25,
 29,
 32,]

如何遍历索引列表并计算每对20s的每个21,22,23的发生率？

到目前为止，我有：

TwentyOnes=0
TwentyTwos=0
TwentyThrees=0

for i in Indexrange:
    for index, row in df.iterrows():
        if index.between(i, i+1):
            if Marker == 21
                Count_of_21s +=
            if Marker == 22
                Count_of_22s +=
            if Marker == 23
                Count_of_23s +=
            else:
                InvalidCount+=

但我正在

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-16-4a72c2a77924> in <module>()
  5 for i in Indexrange:
  6     for index, row in df.iterrows():
----> 7         if index.between(i,i+1):
  8             print(index, row['Marker'])

AttributeError: 'int' object has no attribute 'between'

如何才能获得IndexRange中索引之间的20s /之间的值？

所需的输出为：Counts_of_21s = int，Counts_of_22s = int，Counts_of_23s = int，InvalidCount = int

Answer 1

似乎你需要

df.groupby(df.Marker.eq(20).cumsum()).Marker.value_counts()
Out[1013]: 
Marker  Marker
0       21        1
1       20        1
        21        1
2       2         1
        20        1
        22        1
        23        1
Name: Marker, dtype: int64

更新

df=df.assign(yourid=df.Marker.eq(20).cumsum())
df.loc[(df.yourid<df.yourid.max())&(df.yourid>df.yourid.min())&(df.Marker!=20),:].groupby('yourid').Marker.value_counts()
Out[1021]: 
yourid  Marker
1       21        1
Name: Marker, dtype: int64

Answer 2

这是我的解决方案：

import pandas as pd

csv_df = pd.read_csv('between.txt')

markers = csv_df['Marker'].tolist()
indexrange = csv_df.index[csv_df['Marker'] == 20].tolist()
list_dicts = []

for x in range(len(indexrange)-1):
    currentgroup = {'21': markers[indexrange[x]:indexrange[x+1]].count(21),
                    '22': markers[indexrange[x]:indexrange[x+1]].count(22),
                    '23': markers[indexrange[x]:indexrange[x+1]].count(23)
                    }
    list_dicts.append(currentgroup)

i = 1
for list in list_dicts:
    print(f'Grouping {i}', list)
    i = i+1

温家宝的表现要好得多。

总和行在某些数字之间的索引 - 熊猫Python

2 个答案: