README.md

In zoevanhavre/Zmix: Overfitted Gaussian mixture models with an unknown number of components.

Install Zmix

To install the latest version of the Zmix package from github

library(roxygen2)
library(devtools)
install_github('zoevanhavre/Zmix')
library(Zmix)

Simple Example

Let's generate some data from a mixture of three Gaussian components. Using the Zmix function provided, a Gaussian mixture can be simulated from given the model parameters (\mu_k), (\sigma_k), and mixture weights (\pi_k) for (k=1,\dots, K):

simone<-simudZ(n=100,
               mu=c(-1,4,10),
               sig=c(1,1,2),
               p=c(.2, .7, .1),
               k=3)
#the output includes both the simulated data,
# and each observation's allocation as "Z".
lapply(simone, head)

## $Y
## [1]  3.526882  4.880605 -1.989098  3.404140  2.320297 -1.410440
## 
## $Z
## [1] 2 2 1 2 2 1

Let's run Zmix, fitting 10 components to these data.

run1 <- Zmix_univ_tempered(simone$Y,iter=500,k=10) 

## 
  |                                                                       
  |                                                                 |   0%
  |                                                                       
  |=============                                                    |  20%
  |                                                                       
  |==========================                                       |  40%
  |                                                                       
  |=======================================                          |  60%
  |                                                                       
  |====================================================             |  80%
  |                                                                       
  |=================================================================| 100%

Post processing: main function

pp_run1<-Process_Output_Zmix(run1,Pred_Reps=200, Zswitch_Sensitivity=0.01, isSim=FALSE, Plot_Title="Simone", SaveFileName="Zmix_Run1", Burn=200)

## NULL
##   K0 Probability      MAE      MSE Pmin Pmax Concordance     MAPE     MSPE
## 1  3           1 83.86598 113.2284 0.12 0.84     0.94425 96.54466 149.0588

Result Tables:

pp_run1[[1]]

##   variable factor(k)              value K0
## 1        P         1    0.69(0.57,0.79)  3
## 2       Mu         1    3.98(3.68,4.25)  3
## 3      Sig         1     1.09(0.76,1.6)  3
## 4        P         2    0.18(0.11,0.26)  3
## 5       Mu         2 -1.24(-2.04,-0.51)  3
## 6      Sig         2    1.94(0.92,4.03)  3
## 7        P         3    0.14(0.07,0.23)  3
## 8       Mu         3   9.49(8.06,10.77)  3
## 9      Sig         3    2.38(0.86,6.73)  3

pp_run1[[2]]

##   K0 Probability      MAE      MSE Pmin Pmax Concordance     MAPE     MSPE
## 1  3           1 83.86598 113.2284 0.12 0.84     0.94425 96.54466 149.0588

The plots this makes can be found in the working directory. This includes:

zoevanhavre/Zmix documentation built on May 4, 2019, 11:25 p.m.