组装和分析来自 python 中 dataframe 的列表列表

Question

I've got a.csv file that looks a bit like this:

COL_A    COL_B                   COL_C
1        2020-05-26T00:01:01     99999
2        2020-05-26T00:01:02     99999
3        2020-05-26T00:01:03     99999
4        2020-05-26T00:01:04     2.3
5        2020-05-26T00:01:05     2.3
6        2020-05-26T00:01:06     2.3
7        2020-05-26T00:01:07     99999
8        2020-05-26T00:01:08     99999
9        2020-05-26T00:01:09     3.4
10       2020-05-26T00:01:10     3.4
11       2020-05-26T00:01:11     99999
12       2020-05-26T00:01:12     99999

I'd like to be able to identify the longest continuous span of rows where COL_C is < 5 and return that list of rows. The desired output would be something like:

[
    [4        2020-05-26T00:01:04     2.3,
     5        2020-05-26T00:01:05     2.3,
     6        2020-05-26T00:01:06     2.3]
], 3

The way I have approached this in theory is building a list of lists that meet the criteria, and then using max over the lists with len as the key. I've attempted this:

import pandas as pd

def max_c(csv_file):
    row_list = []
    df = pd.read_csv(csv_file)
    for i, row in df.iterrows():
        while row[2] < 5:
            span = [*row]
            row_list.append(span)
    return max(row_list, key=len)

I know enough to know that this isn't correct syntax for what I'm trying to do and I can even explain why, but do not know enough to get the desired output.

Answer 1

Similar to Quang,find the greater than 5 and create the sub-group, then we just filter out he value more than 5, and get the group with transform count . pick the max count index

s=df.COL_C.ge(5)
s=df.loc[~s,'COL_A'].groupby(s.cumsum()).transform('count')
target=df.loc[s[s==s.max()].index]
Out[299]: 
   COL_A                COL_B  COL_C
3      4  2020-05-26T00:01:04    2.3
4      5  2020-05-26T00:01:05    2.3
5      6  2020-05-26T00:01:06    2.3

Answer 2

I'll use cumsum() to identify blocks and do a groupby:

s = df['COL_C'].lt(5)
sizes = s.groupby([s,(~s).cumsum()]).transform('size') * s

# max block 1 size
# max_size == 0 means all values are >= 5
max_size = sizes.max()


df[sizes==max_size]

Output:

   COL_A                COL_B  COL_C
3      4  2020-05-26T00:01:04    2.3
4      5  2020-05-26T00:01:05    2.3
5      6  2020-05-26T00:01:06    2.3

Details:

s is:

0     False
1     False
2     False
3      True
4      True
5      True
6     False
7     False
8      True
9      True
10    False
11    False
Name: COL_C, dtype: bool

if we just do s.cumsum() then the True obviously belong to different groups. Instead we do (~s).cumsum() we get:

0     1
1     2
2     3
3     3
4     3
5     3
6     4
7     5
8     5
9     5
10    6
11    7
Name: COL_C, dtype: int64

Almost there, but each group of True is now preceded by a row of False . That suggests we group by both s and the negated cumsum.