Forum Discussion
Ignoring spurius peaks
Hi there,
New to this forum so don't know if im in the right place for this sort of question.
I am using Pi add-in in Excel to plot some graphs for pre-defined periods. I am trying to find the maximum value on the trend. However there are often spurius peaks that are generally larger or sharper than the general trend. Does anyone know any techniques/maths/algorithms that can analyse the data and ignore these 'unwanted' peaks and pick the true maximum. I've tried a few things and sometimes it seems to work but other times it doesn't. The shape and form of the trend can be different every time. I've attached an example where there is a peak that may not be a true maximum or is short enough to ignore. I've been looking at how sharp the rise is compared to the previous value and normalising using the average value but this doesnt always produce the value i want. Any suggestions appreciated.