按条件分组 pandas dataframe

Question

I want to cut the continues data into some group. I have some data like this:

Index Age Predict
0     23    0
1     39    0
2     70    0
3     41    1
4     50    0
5     17    0
6     29    1

I try:

df_1 = df[['Age','Predict']]
data = df_1.sort_values(by='Age')

After sorting:

Index Age Predict
5     17    0
0     23    0
6     29    1
1     39    0
3     41    1
2     70    0
4     50    0

What can i do to classifier data into the group:

Index Age Predict

group 1:
5     17    0
0     23    0

group 2:
6     29    1

group 3:
1     39    0

group 4:
3     41    1

group 5:
2     70    0
4     50    0

Thanks for help.

Answer 1

IIUC, the groups you want are created from Predict, where diff between following rows are not equal to 0. so you could create a column:

data_ = df.sort_values('Age')
data_['gr'] = data_['Predict'].diff().ne(0).cumsum()
print (data_)
   Index  Age  Predict  gr
5      5   17        0   1
0      0   23        0   1
6      6   29        1   2
1      1   39        0   3
3      3   41        1   4
4      4   50        0   5
2      2   70        0   5

Or if you want to split your data and not create the group column, one way is to create a dictionary that contains each group

data_ = df.sort_values('Age')
d = {i: dfg 
     for i,(_, dfg) in enumerate(data_.groupby(data_['Predict'].diff().ne(0).cumsum()),1)}
print (d[1])
   Index  Age  Predict
5      5   17        0
0      0   23        0

Answer 2

df.groupby((df['Predict'] != df['Predict'].shift(1)).cumsum())

Basically check if the current value is not the same previous value, if not increment. This will allow you to group by the change in values of Predict

Answer 3

Using .grouby and .cumsum()

for i, grp in data.groupby([(data['Predict'] != data['Predict'].shift()).cumsum()]):
    print('group', i)
    print(grp)

Result:

group 1
   Age  Predict
5   17        0
0   23        0
group 2
   Age  Predict
6   29        1
group 3
   Age  Predict
1   39        0
group 4
   Age  Predict
3   41        1
group 5
   Age  Predict
4   50        0
2   70        0