Calculate Difference between columns where criteria in another column is met

Question

I have a dataset as follows:

import pandas as pd
df = pd.DataFrame(
    {"Time": [1,2,3,1,2,3],
     "Prop1": ["A","B","A","A","B","B"],
     "Prop2": [4, 5, 1, 5, 4, 3]})

Time	Prop1	Prop2
1	A	4
2	B	5
3	A	1
1	A	5
2	B	4
3	B	3

What I want is the differences...

1. At each time step
2. For Prop1 value
3. between Prop2

From this dataset I'd like to receive something like...

Time	Prop1	Prop2	Prop2Diff
1	A	4	NaN
2	B	5	NaN
3	A	1	NaN
1	A	5	(5-4)= 1
2	B	4	(4-5)= -1
3	B	3	NaN

I know there is .diff() function, but I am uncertain how to apply the type of conditions I'm wanting.

This is in an effort to remove deep loops from my scripts, I have already done:

time = df["Time"].unique()
for cTime in time:
    for prop1 in df[df["Time"] == cTime,"Prop1"].unique():
        df.loc[(df["Time"]==cTime) & (df["Prop1"] == prop1] = cDF["Prop2"][1] - cDF["Prop2"][0]

And other variations of the loops.

This is operating on very large datasets, so every loop I can remove tremendously helps.

Answer 1

Code

df['Prop2Diff'] = df.groupby(['Time', 'Prop1'])['Prop2'].diff()

df

   Time Prop1  Prop2  Prop2Diff
0     1     A      4        NaN
1     2     B      5        NaN
2     3     A      1        NaN
3     1     A      5        1.0
4     2     B      4       -1.0
5     3     B      3        NaN