I have this data frame and I want to select 10 rows before and after on a specific column. I have reached up to this point but I was wondering how to make it more elegant in a lambda python expression as I need to run this on a loop 10 thousand times.
import pandas as pd
df = pd.DataFrame(data=np.random.rand(90),
index=pd.date_range('2015-01-01','2015-03-31'),columns=['A'])
I have reached to this as an solution in progress:
10 observations before and after:
df.loc['2015-01-17':].head(11)[1:11].transpose() ! before
df.loc[:'2015-01-17'].tail(11)[0:10].transpose() ! after
So, how can I make this is in a loop with a lambda expression and having not only one index but two indexes?
Really simple using index.get_loc. Get the index of the label, and slice accordingly.
idx = df.index.get_loc('2015-01-17')
df.iloc[idx - 10 : idx + 10]
A
2015-01-07 0.262086
2015-01-08 0.836742
2015-01-09 0.094763
2015-01-10 0.133500
2015-01-11 0.285372
2015-01-12 0.338112
2015-01-13 0.451852
2015-01-14 0.163001
2015-01-15 0.247186
2015-01-16 0.227053
2015-01-17 0.837647
2015-01-18 0.918334
2015-01-19 0.514731
2015-01-20 0.207688
2015-01-21 0.700314
2015-01-22 0.363784
2015-01-23 0.811346
2015-01-24 0.079030
2015-01-25 0.051900
2015-01-26 0.520310