How to add values one by one in a new column with pandas

Question

I have a column with acceleration values, and I’m trying to integrate them in a new column. Here’s what I want as output :

  A B
0 a b-a
1 b c-b
2 c d-c
3 d …-d
…

I’m currently doing like that

l=[]
for i in range(len(df)):
   l.append(df.values[i+1][0]-df.values[i][0])
df[1]=l

That’s very slow to process. I have over a million lines, and this in 20 different csv files. Is there a way to do it faster ?

I'm sure that's possible with vectorized operations. Can't say anything else given the information you provided. — timgeb, May 05 '22 at 08:23
@timgeb dont really know what else to add, i think my question is clear, and I couldn’t find any answer or help anywhere — bappi, May 05 '22 at 08:25
Your question has no sample data, no code, no actual output and no desired output. It could be much, much clearer. — timgeb, May 05 '22 at 08:26
See [How to make good reproducible pandas examples](https://stackoverflow.com/a/20159305/3620003). — timgeb, May 05 '22 at 08:28

score 0 · Accepted Answer · answered May 05 '22 at 08:47

0

IIUC, you can use diff:

df = pd.DataFrame({'A': [0,2,1,10]})
df['B'] = -df['A'].diff(-1)

output:

answered May 05 '22 at 08:47

mozway

note that `df['A'].diff().shift(1)` also works but is slightly less efficient – mozway May 05 '22 at 09:12

1 Answers1