如何迭代 Pandas 数据框中的多列？

Question

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.

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. 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. Something like this:

--------------------------------
  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. 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. Similarly, I am then comparing each ID value to the next row (as this will denote the end of speech).

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'])

Answer 1

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:

   start  end
0      4    7
1     12   15

Answer 2

Use the following code:

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:

   start  end
0      4    7
1     12   15