[Python] Expanding/Modifying a data frame by adding data points based on difference between...

I have a data frame, which is the annotation of animal behavior recording frame by frame. But due to the limitation of the annotation software, the exported data frame is like this:

Frame stop Behavior
Frame start
0 2 turning
2 5 immobility
5 9 mild movement
9 11 immobility
16 20 jump


Code:

data = {
'Frame start': [0, 2, 5, 9, 16],
'Frame stop': [2, 5, 9, 11, 20],
'Behavior': ['turning', 'immobility', 'mild movement', 'immobility', 'jump']}

df = pd.DataFrame(data)
df.set_index('Frame start', inplace=True)


I want to modify the data frame to be like this:

Frame start
0 turning
1 turning
2 immobility
3 immobility
4 immobility
5 mild movement
6 mild movement
7 mild movement
8 mild movement
9 immobility
10 immobility
11 NaN
12 NaN
13 NaN
14 NaN
15 NaN
16 jump
17 jump
18 jump
19 jump


How can I do that?

Since I did not annotate every frame of a video, there are gaps of frames with no values, i.e. between index 4 and 5, frame 11-15 in this example. Therefore, I cannot re-index the data frame using the "frame start" column and simply <ffill> to fill up the missing values, which will look like the following:

new_index = [*range(0, 20)]
new_df = df.reindex(new_index, method='ffill')
final_df = new_df.loc[:, 'Behavior']

print(final_df)

Frame start
0 turning
1 turning
2 immobility
3 immobility
4 immobility
5 mild movement
6 mild movement
7 mild movement
8 mild movement
9 immobility
10 immobility
11 immobility
12 immobility
13 immobility
14 immobility
15 immobility
16 jump
17 jump
18 jump
19 jump

Continue reading...