[英]How to iterate multiple columns in a Pandas dataframe?
I have the following dataframe where one column denotes the ID (0,1) of a speaker for each second of a conversation, and the other column denotes the seconds passed in that conversation.我有以下数据框,其中一列表示对话中每一秒的说话者 ID (0,1),另一列表示该对话中经过的秒数。
myDF = pd.DataFrame({'ID': [0,0,0,0,1,1,1,1,0,0,0,0,1,1,1,1], 'seconds': (np.arange(16))})
-------------------------------
ID Seconds
-------------------------------
0 0
0 1
0 2
0 3
1 4
1 5
1 6
1 7
0 8
0 9
0 10
0 11
1 12
1 13
1 14
1 15
-------------------------------
I am only interested in speaker ID 1, where hopefully it is clear that the boundaries of speaker 1 speech is between seconds 4-7, and between 12-15.我只对演讲者 ID 1 感兴趣,希望演讲者 1 演讲的边界在 4-7 秒和 12-15 秒之间很明显。 What I want to generate is a separate dataframe that contains the start and end of each speaker 1 speech segment, where each row is an uninterrupted period of speech.
我想要生成的是一个单独的数据帧,其中包含每个说话者 1 语音段的开始和结束,其中每一行是一段不间断的语音。 Something like this:
像这样的东西:
--------------------------------
start end
--------------------------------
4 7
12 15
--------------------------------
I have some non-working pseudo-code that hopefully outlines what I am trying to achieve, but as yet I cannot find the right solution.我有一些非工作的伪代码,希望能概述我想要实现的目标,但到目前为止我还找不到正确的解决方案。 In essence, for each row I am comparing the ID value with the previous row (because a change in ID denotes the start of speech) and adding the corresponding seconds value to the bdry dataframe.
实质上,对于每一行,我将 ID 值与前一行进行比较(因为 ID 的变化表示语音的开始)并将相应的秒值添加到 bdry 数据帧。 Similarly, I am then comparing each ID value to the next row (as this will denote the end of speech).
同样,然后我将每个 ID 值与下一行进行比较(因为这将表示语音结束)。
bdry = pd.DataFrame(columns=['start','end'])
for i in myDF:
if i['ID'] == 1:
if i.ID != i['ID'].shift(): # compare ID with previous
bdry['start'].append(i['seconds'])
if i.ID != i['ID'].shift(-1): # compare ID with next
bdry['end'].append(i['seconds'])
import pandas as pd
from itertools import groupby
myDF = pd.DataFrame({'ID': [0,0,0,0,1,1,1,1,0,0,0,0,1,1,1,1], 'seconds': (np.arange(16))})
tmp, m = [], myDF['ID'] == 1
for v, g in groupby(zip(m.index, m), lambda k: k[1]):
if v:
g = list(g)
tmp.append((g[0][0], g[-1][0]))
df = pd.DataFrame(tmp, columns=['start', 'end'])
print(df)
Prints:印刷:
start end
0 4 7
1 12 15
Use the following code:使用以下代码:
result = myDF.groupby((myDF.ID != myDF.ID.shift()).cumsum()).agg(
ID=('ID', 'first'), start=('seconds', 'first'), end=('seconds', 'last'))\
.query('ID == 1').drop(columns='ID').reset_index(drop=True)
For your data sample the result is:对于您的数据样本,结果是:
start end
0 4 7
1 12 15
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.