Reputation: 395
I have the following list:
a = [80, 79, 30, 15, 12, 14, 20, 15, 10, 45, 52, 59, 51, 60, 72, 77, 60, 15, 20, 10, 12]
I would like to get a list 'b' with the same length as 'a' where each index shows a Centred Moving Window 10th percentile of +-3 locations of each index.
So I am interested in a list that shows the following:
b[0] = np.percentile(a[0:2], 10)
b[1] = np.percentile(a[0:3], 10)
b[2] = np.percentile(a[0:4], 10)
b[3] = np.percentile(a[0:5], 10)
b[4] = np.percentile(a[1:6], 10)
b[5] = np.percentile(a[2:7], 10)
.....
Thanks
Upvotes: 1
Views: 76
Reputation: 323266
Convert your list to dataframe
df=pd.DataFrame(a).reset_index()
Then we using apply
df['index'].apply(lambda x : np.percentile(df[0].iloc[np.clip(x-3,0,None):x+2],10))
Out[1429]:
0 79.1
1 39.8
2 19.5
3 13.2
4 12.8
5 12.8
6 12.8
7 10.8
8 11.6
9 12.0
10 12.0
11 24.0
12 47.4
13 51.4
14 54.2
15 54.6
16 33.0
17 17.0
18 12.0
19 10.8
20 10.6
Name: index, dtype: float64
Upvotes: 1