Streaming peak detection based on https://stackoverflow.com/questions/22583391/peak-signal-detection-in-realtime-timeseries-data/22640362#22640362

streaming_peak_detection.py

from window_avg_std import window_avg_std


def peak_detection(window, threshold, influence):
    """ Smoothed z-score algo
    https://stackoverflow.com/questions/22583391/peak-signal-detection-in-realtime-timeseries-data/43512887#43512887 """
    elem = yield  # get first elem
    prev_elem = elem  # previous elem (influenced)
    elem_index = 0
    prev_avg = elem
    prev_std = 0
    stat = window_avg_std(window)
    next(stat)  # roll generator to first yeild
    while True:
        # print(f'elem: {elem}, prev_avg: {prev_avg}, prev_std: {prev_std}, index: {elem_index}')
        if abs(elem - prev_avg) > threshold * prev_std and \
                elem_index >= window-1:  # detection starts after first window'th elements
            signal = 1 if elem > prev_avg else -1
            elem = influence * elem + (1-influence) * prev_elem  # smooth peak
        else:
            signal = 0
        prev_avg, prev_std = stat.send(elem)
        prev_elem = elem
        elem_index += 1
        elem = yield signal


if __name__ == '__main__':
    data = [1, 1, 1.1, 1, 0.9, 1, 1, 1.1, 1, 0.9, 1, 1.1, 1, 1, 0.9, 1, 1, 1.1, 1, 1, 1, 1, 1.1, 0.9, 1, 1.1, 1, 1, 0.9,
            1, 1.1, 1, 1, 1.1, 1, 0.8, 0.9, 1, 1.2, 0.9, 1, 1, 1.1, 1.2, 1, 1.5, 1, 3, 2, 5, 3, 2, 1, 1, 1, 0.9, 1, 1, 3,
            2.6, 4, 3, 3.2, 2, 1, 1, 0.8, 4, 4, 2, 2.5, 1, 1, 1]
    detector = peak_detection(window=3, threshold=3, influence=0.2)
    next(detector)
    while data:
        elem = data.pop()
        signal = detector.send(elem)
        print('elem: {}, signal: {}'.format(elem, signal))

window_avg_std.py

from queue import Queue


def window_avg_std(window=3):
    """ streaming calc avg and std """
    buf = Queue()
    elem_sum1 = 0
    elem_sum2 = 0  # sum of squared
    elem = yield
    while True:
        buf.put_nowait(elem)
        elem_sum1 += elem
        elem_sum2 += elem * elem
        if buf.qsize() > window:
            x = buf.get_nowait()
            elem_sum1 -= x
            elem_sum2 -= x * x
        avg = elem_sum1 / buf.qsize()
        std = (elem_sum2 - elem_sum1) / buf.qsize()
        elem = yield avg, std


if __name__ == '__main__':
    f = window_avg_std()
    next(f)
    for i in range(1, 7):
        v = f.send(i)
        print('i={} v={}'.format(i, v))

Need corrections for std!