如何通过 groupby 计算 pandas dataframe 中的更改数量？

Question

I have a data that looks like:

df = pd.DataFrame({
    'ID': [1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2],
    'DATE': ['1/1/2015','1/2/2015', '1/3/2015','1/4/2015','1/5/2015','1/6/2015','1/7/2015','1/8/2015',
             '1/9/2016','1/2/2015','1/3/2015','1/4/2015','1/5/2015','1/6/2015','1/7/2015'],
    'CD': ['A','A','A','A','B','B','A','A','C','A','A','A','A','A','A']})

I would like to count # of changes that occurs by ID and CD. How can I get the desired result. When I tried cumcount, it will count same groupby and give it different numbers.

What I get is:

What I am expecting is:

Answer 1

your count column in desired output means group

First

make grouper to divide group (changed bool to int for ease of viewing)

col = ['ID', 'CD']
grouper = df[col].ne(df[col].shift(1)).any(axis=1).astype('int')

grouper

0     1
1     0
2     0
3     0
4     1
5     0
6     1
7     0
8     1
9     1
10    0
11    0
12    0
13    0
14    0
dtype: int32

Second

divide group in same ID (I made grouper to count column because had to create count column anyway.)

df.assign(count=grouper).groupby('ID')['count'].cumsum()

output:

0     1
1     1
2     1
3     1
4     2
5     2
6     3
7     3
8     4
9     1
10    1
11    1
12    1
13    1
14    1
Name: count, dtype: int32

Last

make output to count column

df.assign(count=df.assign(count=grouper).groupby('ID')['count'].cumsum())

result:

    ID  DATE       CD   count
0   1   1/1/2015    A   1
1   1   1/2/2015    A   1
2   1   1/3/2015    A   1
3   1   1/4/2015    A   1
4   1   1/5/2015    B   2
5   1   1/6/2015    B   2
6   1   1/7/2015    A   3
7   1   1/8/2015    A   3
8   1   1/9/2016    C   4
9   2   1/2/2015    A   1
10  2   1/3/2015    A   1
11  2   1/4/2015    A   1
12  2   1/5/2015    A   1
13  2   1/6/2015    A   1
14  2   1/7/2015    A   1

---

Update full code

more simple full code with advice of @cottontail

col = ['ID', 'CD']
grouper = df[col].ne(df[col].shift(1)).any(axis=1).astype('int')
df.assign(count=grouper.groupby(df['ID']).cumsum())

Answer 2

Lets group on ID column and use shift on CD to check for changes then use cumsum to create sequential counter

df['count'] = df.groupby('ID')['CD'].apply(lambda s: s.ne(s.shift()).cumsum())

Result

    ID      DATE CD  count
0    1  1/1/2015  A      1
1    1  1/2/2015  A      1
2    1  1/3/2015  A      1
3    1  1/4/2015  A      1
4    1  1/5/2015  B      2
5    1  1/6/2015  B      2
6    1  1/7/2015  A      3
7    1  1/8/2015  A      3
8    1  1/9/2016  C      4
9    2  1/2/2015  A      1
10   2  1/3/2015  A      1
11   2  1/4/2015  A      1
12   2  1/5/2015  A      1
13   2  1/6/2015  A      1
14   2  1/7/2015  A      1