Smoothing online data on Python with savgol_filter from scipy.signal library

Question

I'd like to filter online data with savgol_filter from scipy.signal library. But when I was trying to use it for online data (when new elements appear one by one) I realized that savgol_filter works with online data with some delay (window_length//2) in comparison to how it works with offline data (their elements are available for calculation all at once). I use code similar to that (see below please)

from queue import Queue, Empty
import numpy as np
from scipy.signal import savgol_filter

window_size = 5
data = list()
q = Queue()
d = [2.22, 2.22, 5.55, 2.22, 1.11, 0.01, 1.11, 4.44, 9.99, 1.11, 3.33]
for i in d:
    q.put(i)

res = list()
while not q.empty():
    element = q.get()
    data.append(element)
    length = len(data)
    npd = np.array(data[length - window_size:])
    if length >= window_size:
        res.append(savgol_filter(npd , window_size, 2)[window_size // 2])

npd = np.array(data)
res2 = savgol_filter(npd , window_size, 2)

np.set_printoptions(precision=2)
print('source data ', npd)
print('online res  ', np.array(res))
print('offline res ', res2)

1. Am I right in my assumption? Can it be corrected somehow?
2. If I am right could you please advice similar filter with no such issue in calculations?

Answer 1

It looks like Kalman filters family are doing what I expect. This is because they are optimal in terms of "Mean square error". Implementation can be found here for example.