algorithm - Best way to extract the minimum and maximum value of the dense part of a distribution? -
i have distributions of point distance parallel line. each distributions have area more populated represent point channel. extract minimum , maximum represented red line in graphs? eyes can how robustly algorithm?
the x axis represents perpendicular distance of points line 0 100m.
the y axis represents number of points have distance in bin.
example 1

example 2

since distribution comes set of distances points line, , values in order, may try compute normal distribution models samples. there, margins (your red bars) mean +/- x*sigma, x can value want (maybe 1 or 2).
if points not in order, may percentile (0.25, example) of full list of values threshold, , assume populated part of distribution starts there values higher percentile.
Comments
Post a Comment