Series data increasing numbers

Question

In Pandas Dataframe, I input data unique key.

For Example Inputs:

time                                  range                   
2018-03-04 00:00:06.520                 0                     
2018-03-04 00:00:07.130                 0                      
2018-03-04 00:00:07.850                 1                      
2018-03-04 00:00:08.420                 1                      
2018-03-04 00:00:09.210                 2                      
2018-03-04 00:00:10.070                 2                      
2018-03-04 00:00:10.840                 3                      
2018-03-04 00:00:11.230                 3                      
2018-03-04 00:00:11.980                 4                      
2018-03-04 00:00:12.560                 4                      
2018-03-04 00:00:13.120                 0                      
2018-03-04 00:00:13.790                 0                      
2018-03-04 00:00:14.330                 1                      
2018-03-04 00:00:15.280                 1                      
2018-03-04 00:00:15.960                 2                      
2018-03-04 00:00:16.420                 2                      
2018-03-04 00:00:17.090                 3      

I wanna Output dataFrame is here.

time                                  range                   Key
2018-03-04 00:00:06.520                 0                      1
2018-03-04 00:00:07.130                 0                      1
2018-03-04 00:00:07.850                 1                      1
2018-03-04 00:00:08.420                 1                      1
2018-03-04 00:00:09.210                 2                      1
2018-03-04 00:00:10.070                 2                      1
2018-03-04 00:00:10.840                 3                      1
2018-03-04 00:00:11.230                 3                      1
2018-03-04 00:00:11.980                 4                      1
2018-03-04 00:00:12.560                 4                      1
2018-03-04 00:00:13.120                 0                      2
2018-03-04 00:00:13.790                 0                      2
2018-03-04 00:00:14.330                 1                      2
2018-03-04 00:00:15.280                 1                      2
2018-03-04 00:00:15.960                 2                      2
2018-03-04 00:00:16.420                 2                      2
2018-03-04 00:00:17.090                 3                      2
...

I wanna using range & Time get Key Values that increasing..

How can I do it?

Answer 1

diff

df.assign(Key=df.range.diff().lt(0).cumsum().add(1))

                      time  range  Key
0  2018-03-04 00:00:06.520      0    1
1  2018-03-04 00:00:07.130      0    1
2  2018-03-04 00:00:07.850      1    1
3  2018-03-04 00:00:08.420      1    1
4  2018-03-04 00:00:09.210      2    1
5  2018-03-04 00:00:10.070      2    1
6  2018-03-04 00:00:10.840      3    1
7  2018-03-04 00:00:11.230      3    1
8  2018-03-04 00:00:11.980      4    1
9  2018-03-04 00:00:12.560      4    1
10 2018-03-04 00:00:13.120      0    2
11 2018-03-04 00:00:13.790      0    2
12 2018-03-04 00:00:14.330      1    2
13 2018-03-04 00:00:15.280      1    2
14 2018-03-04 00:00:15.960      2    2
15 2018-03-04 00:00:16.420      2    2
16 2018-03-04 00:00:17.090      3    2

---

flatnonzero and repeat

a = np.diff(np.flatnonzero(np.concatenate(
    [[True], np.diff(df.range.values) < 0, [True]]
)))

df.assign(Key=np.arange(a.size).repeat(a) + 1)

                      time  range  Key
0  2018-03-04 00:00:06.520      0    1
1  2018-03-04 00:00:07.130      0    1
2  2018-03-04 00:00:07.850      1    1
3  2018-03-04 00:00:08.420      1    1
4  2018-03-04 00:00:09.210      2    1
5  2018-03-04 00:00:10.070      2    1
6  2018-03-04 00:00:10.840      3    1
7  2018-03-04 00:00:11.230      3    1
8  2018-03-04 00:00:11.980      4    1
9  2018-03-04 00:00:12.560      4    1
10 2018-03-04 00:00:13.120      0    2
11 2018-03-04 00:00:13.790      0    2
12 2018-03-04 00:00:14.330      1    2
13 2018-03-04 00:00:15.280      1    2
14 2018-03-04 00:00:15.960      2    2
15 2018-03-04 00:00:16.420      2    2
16 2018-03-04 00:00:17.090      3    2

Answer 2

I think you can use lt() (less than), shift(), and cumsum(). Together, you can make these cumulatively count each time the column range stops increasing (i.e. when the range value is less than the previous range value).

df['Key'] = df['range'].lt(df['range'].shift()).cumsum() + 1

>>> df
                       time  range  Key
0   2018-03-04 00:00:06.520      0    1
1   2018-03-04 00:00:07.130      0    1
2   2018-03-04 00:00:07.850      1    1
3   2018-03-04 00:00:08.420      1    1
4   2018-03-04 00:00:09.210      2    1
5   2018-03-04 00:00:10.070      2    1
6   2018-03-04 00:00:10.840      3    1
7   2018-03-04 00:00:11.230      3    1
8   2018-03-04 00:00:11.980      4    1
9   2018-03-04 00:00:12.560      4    1
10  2018-03-04 00:00:13.120      0    2
11  2018-03-04 00:00:13.790      0    2
12  2018-03-04 00:00:14.330      1    2
13  2018-03-04 00:00:15.280      1    2
14  2018-03-04 00:00:15.960      2    2
15  2018-03-04 00:00:16.420      2    2
16  2018-03-04 00:00:17.090      3    2