statistics - Scipy Guassian_kde Nomalisation -


i've been using scipy.stats.gausian_kde have few questions output. i've plotted normalised histogram , gaussian_kde plot on same graph. why y-values vastly different? understanding gaussian_kde plot should touch tips of histograms, roughly. using scipy.integrate.quad functions determined area under graph 0.7, rather 1.0, expected.

actually want gaussian_kde represent non-normalised histogram, know how can that? graph here

your expectations little off. area under each of kde's peaks should equal area in corresponding bars. appears hold, eye. nonadaptive kdes global bandwidth estimate (like scipy.stats.gaussian_kde) tend broaden multimodal distributions sharp peaks.

as underestimate of total area under kde, cannot without data , code used integration.

in order make kde approximate unnormalized histogram, need multiply (bin_width*n) n total number of data points.


Comments

Popular posts from this blog

Why does Ruby on Rails generate add a blank line to the end of a file? -

keyboard - Smiles and long press feature in Android -

node.js - Bad Request - node js ajax post -