I have the following pandas DataFrame called df:
timestamp param_1 param_2
0.000 -0.027655 0.0
0.25 -0.034012 0.0
0.50 -0.040369 0.0
0.75 -0.046725 0.0
1.00 -0.050023 0.0
1.25 -0.011015 0.0
1.50 -0.041366 0.0
1.75 -0.056723 0.0
2.00 -0.013081 0.0
Now I need to add two new columns created from the following lists:
timestamp_new = [0.5, 1.0, 1.5, 2.0]
param_3 = [10.0, 25.0, 15.0, 22.0]
The problem is that timestamp_new has a different granularity. Thus, I need to interpolate (linearly) both timestamp_new and param_3 in order to fit the granularity of timestamp in df.
Expected result (please notice that I interpolated param_3 values randomly just to show the format of an expected result):
timestamp param_1 param_2 param_3
0.000 -0.027655 0.0 8.0
0.25 -0.034012 0.0 9.0
0.50 -0.040369 0.0 10.0
0.75 -0.046725 0.0 20.0
1.00 -0.050023 0.0 25.0
1.25 -0.011015 0.0 18.0
1.50 -0.041366 0.0 15.0
1.75 -0.056723 0.0 17.0
2.00 -0.013081 0.0 22.0
Is there any way to do it?
Let's try reindex().interpolate:
ref_df = pd.Series(param_3, index=timestamp_new)
new_vals = (ref_df.reindex(df['timestamp'])
.interpolate('index')
.bfill() # fill the first few nans
.ffill() # fill the last few nans
)
df['param_3'] = df['timestamp'].map(new_vals)
Output:
timestamp param_1 param_2 param_3
0 0.00 -0.027655 0.0 10.0
1 0.25 -0.034012 0.0 10.0
2 0.50 -0.040369 0.0 10.0
3 0.75 -0.046725 0.0 17.5
4 1.00 -0.050023 0.0 25.0
5 1.25 -0.011015 0.0 20.0
6 1.50 -0.041366 0.0 15.0
7 1.75 -0.056723 0.0 18.5
8 2.00 -0.013081 0.0 22.0