Fig. 8.6 and Tab. 8.2: Example for EM
szcf-weiya opened this issue · 2 comments
szcf-weiya commented
it is summarized in https://esl.hohoweiya.xyz/notes/Mixture-Gaussian/index.html#em
szcf-weiya commented
in R
pi = 0.554590232459113
Finish!
mu1 = 4.65591280772299
mu2 = 1.0831618075166
sigma1^2 = 0.818793570810866
sigma2^2 = 0.811370618556585
the estimations are (slightly) different the original.
szcf-weiya commented
in Julia
a variant version that assumes the ratio of sigma1/sigma2 is also provided
julia> μ1
1.0831617392634196
julia> μ2
4.655912729215212
julia> σ1^2
0.8113705078083597
julia> σ2^2
0.8187936852297709
julia> w
0.5545902120768698which is quite close to the above calculated by R.
| pi | log-likelihood |
|---|---|
 |
 |