Running Mean of Circular Data: Average and Standard Deviation?

Question

Given such a list:

mylist = [1, 2, 3, 4, 5, 6, 7]

with N = 3 as the size of running mean at each step.

What is the fastest way to calculate the average and standard deviation of this list?

If it was only average np.convolve could do the job, but what about the standard deviation? or standard error?

schlodinger · Answer 1 · 2020-02-15T17:31:32.017

1

import numpy as np

mylist = [1, 2, 3, 4, 5, 6, 7]
double = mylist * 2
N = 3
mean_std = [(np.mean(double[i:i+N]), np.std(double[i:i+N])) for i in range(len(mylist))]

edited Feb 15 '20 at 17:31

answered Feb 15 '20 at 15:56

schlodinger

537
3
14

This is correct with not circular data, as it gives only 6 values. How does it work with circular data, i.e. 7 values? – Vigor831 Feb 15 '20 at 16:15
what do you mean by circular data? – schlodinger Feb 15 '20 at 16:16
mylist is circular, meaning that I should also compute the average also for [6,7,1], [7,1,2], getting an average with 7 values – Vigor831 Feb 15 '20 at 17:00
I modified the solution so that it works for your case – schlodinger Feb 15 '20 at 17:31

Grzegorz Skibinski · Accepted Answer · 2020-02-15T17:03:31.543

1

Try:

import numpy as np
N=3
mylist = [1, 2, 3, 4, 5, 6, 7]

res=np.vstack([mylist[i:]+mylist[:i] for i in range(N)])

ma=res.mean(axis=0)
std=res.std(axis=0)

Just for moving average you can do: https://stackoverflow.com/a/14314054/11610186

edited Feb 15 '20 at 17:03

answered Feb 15 '20 at 16:28

Grzegorz Skibinski

12,624
2
11
34

This is correct for non-circulat data, what about circular ones? I also need to compute the average for [6,7,1] and [7,1,2] – Vigor831 Feb 15 '20 at 17:00

Running Mean of Circular Data: Average and Standard Deviation?

2 Answers2