The way PeakDetector is designed makes it a bit (and unnecessarily) cumbersome and counter-intuitive. One needs to now about the normal law and outliers. Instead, we could simply be able to send a signal directly to the PeakDetector which could be in charge of detecting peaks.
Furthermore, instead of using the principle of "outliers" it would be interesting to use a concept of "sensitvity" as a percentage (%) (0..1) For this we could use the "logit" function and calculate the n-stddev from a probability.
The way PeakDetector is designed makes it a bit (and unnecessarily) cumbersome and counter-intuitive. One needs to now about the normal law and outliers. Instead, we could simply be able to send a signal directly to the PeakDetector which could be in charge of detecting peaks.
Furthermore, instead of using the principle of "outliers" it would be interesting to use a concept of "sensitvity" as a percentage (%) (0..1) For this we could use the "logit" function and calculate the n-stddev from a probability.