参考 pandas dataframe 中的列值减去固定行值

Question

I would like to subtract a fixed row value in rows, in reference to their values in another column.

My data looks like this:

       TRACK    TIME  POSITION_X
0        1        0       12       
1        1        30      13       
2        1        60      15       
3        1        90      11      
4        2         0      10     
5        2        20      11
6        2        60      13
7        2        90      17

I would like to subtract a fixed row value (WHEN TIME=0) of the POSITION_X column in reference to the TRACK column, and create a new column ("NEW_POSX") with those values. The output should be like this:

       TRACK    TIME  POSITION_X   NEW_POSX
0        1        0       12         0   
1        1        30      13         1
2        1        60      15         3
3        1        90      11        -1
4        2         0      10         0
5        2        20      11         1
6        2        60      13         3
7        2        90      17         7

I have been using the following code to get this done:

import pandas as pd

data = {'TRACK':  [1,1,1,1,2,2,2,2],
        'TIME': [0,30,60,90,0,20,60,90],
        'POSITION_X': [12,13,15,11,10,11,13,17],
        }

df = pd.DataFrame (data, columns = ['TRACK','TIME','POSITION_X'])
df['NEW_POSX']= df.groupby('TRACK')['POSITION_X'].diff().fillna(0).astype(int)
df.head(8)

... but I don't get the desired output. Instead, I get a new column where every row is subtracted by the previous row (according to the "TRACK" column):

       TRACK    TIME  POSITION_X   NEW_POSX
0        1        0       12         0   
1        1        30      13         1
2        1        60      15         2
3        1        90      11        -4
4        2         0      10         0
5        2        20      11         1
6        2        60      13         2
7        2        90      17         4

can anyone help me with this?

Answer 1

You can use transform and first to get the value at time 0, and then substract it to the 'POSITION_X' column:

s=df.groupby('TRACK')['POSITION_X'].transform('first')
df['NEW_POSX']=df['POSITION_X']-s

#Same as: 
#df['NEW_POSX']=df['POSITION_X'].sub(s)

Output:

df
   TRACK  TIME  POSITION_X  NEW_POSX
0      1     0          12         0
1      1    30          13         1
2      1    60          15         3
3      1    90          11        -1
4      2     0          10         0
5      2    20          11         1
6      2    60          13         3
7      2    90          17         7