Threshold normalization for function indexes
Issue #8
resolved
Hi, thank you for your excellent work.
When I was using the function peak.indexes, I was confused by the parameter "threshold" for a while until I read the source code. Currently, the threshold is normalized by
thres_normalized = thres / (max(y) - min(y))
but not every possible value of threshold could be represented between 0 and 1 in this way. For example:
y = [90, 95, 100]
if I want a threshold equal to 98, my parameter should be 9.8
Maybe the normalization should be like this:
thres_normalized = (thres - min(y)) / (max(y) - min(y))
So in the implementation of the function, it's like
def indexes(y, thres=0.3, min_dist=1):
...
thres = thres * (np.max(y) - np.min(y)) + np.min(y)
...
Shiyu
Comments (2)
-
repo owner -
repo owner - changed status to resolved
Fixes issue
#8(wrong normalization of threshold)→ <<cset 9d11b4b28557>>
- Log in to comment
Hi,
Thanks for pointing this issue. You are absolutely correct!