findInt: Finds the intersection between the two fitted Gaussians.

In MarcelaCespedes/FSPmix: Exploratory tool for fast unsupervised features selection and clustering using mixture models

Description

Find the real value intersection (deepest part of the valley) between the means of two univariate Guassian distributions

Usage

findInt(mu1, mu2, sd1, sd2, p1, p2, filter=TRUE)

Arguments

mu1, mu2	fitted means of two components
sd1, sd2	fitted standard deviations of two components
p1, p2	fitted proportions for the two components
filter	boolean: if TRUE, try to find the "middle" & real-valued intersection

Value

rr	Real value of the intersection between the means of the two Guassian distributions
sw	boolean: value 0 if real continuous intersection value was found; otherwise if no real (and imaginary) value was found sw = 0

Author(s)

Amy Chan

References

Similar to this (but with mixing proportions) https://stats.stackexchange.com/questions/311592/how-to-find-the-point-where-two-normal-distributions-intersect

Examples

library(mixtools)
y <- rnormmix(1e3, lambda=c(0.2, 0.8), mu=c(-0.5, 0.5), sigma=c(0.1, 0.3))
hist(y)
m <- normalmixEM(y, k=2)
thr <- findInt(m$mu[1], m$mu[2], m$sigma[1], m$sigma[2], m$lambda[1], m$lambda[2])
thr

MarcelaCespedes/FSPmix documentation built on May 12, 2020, 5:49 p.m.