Original link: http://tecdat.cn/?p=23825
brief introduction
This paper introduces the implementation of finite normal mixture model in R software for model-based clustering, classification and density estimation. The EM algorithm is used to estimate the parameters of normal mixed models with various covariance structures, and the simulation functions based on these models are provided. In addition, it also includes the function of combining model-based hierarchical clustering, EM of mixed distribution estimation and Bayesian information criterion (BIC), and a comprehensive strategy for clustering, density estimation and discriminant analysis. Other functions can be used to display and visualize fitting models as well as clustering, classification and density estimation results.
clustering
head(X)pairs(X)
plot(BIC)
summary(BIC)
summary(mod1, parameters = TRUE) 
plot(mod1)
table(class, classification)plot(mod1, what = "uncertainty")
clustICL(X)
summary(ICL)
BootstrapLRT(X)initialization
EM algorithm is used for maximum likelihood estimation. The initialization of EM is performed using partitions obtained from the cluster hierarchy clustering.
hclust(X, use = "SVD"))
Clustbic (x, initialization)) \ 
hc2
clustBIC(X, initialization ) 
hclust(X, model= "EEE")) 
summary(BIC3) 
Update BIC by merging best results.
BIC(BIC1, BIC2, BIC3)
Univariate fitting using random starting points is obtained by creating random agglomerations and merging best results.
for(j in 1:20)
{
rBIC <- mclustBIC(
initi ))
BIC <- update(BIC, rBIC)
}
clust(ga, BIC)
classification
EDDA
X <- iris\[,1:4\]
head(X)
clustDA(X, class, "EDDA")
plot(mod2)
MclustDA
table(class)
head(X)
clustDA(X, class)
plot(mod3, 2) 
plot(mod3, 3)
Cross validation error
cv(mod2, nfold = 10)
unlist(cv\[3:4\]) 
cv(mod3, nf = 10)
unlist(cv\[3:4\])
density estimation
Univariate
clust(acid)
plot(mod4, "BIC")
plot(mod4, "density", acidity)
plot(mod4, "diagnostic", "cdf")
Multivariable
clu(faithful)
summary(mod5)
plot(mod5, "BIC")
plot(mod5, "density",faithful)
Bootstrap inference
summary(boot1, what = "se") 
summary(boot1, what = "ci") 
summary(boot4, what = "se")
plot(boot4)
Dimensionality reduction
clustering
plot(mod1dr, "pairs")
plot(mod1dr)
plot(mod1dr, "scatterplot")
plot(mod1dr)
classification
summary(mod2dr)
plot(mod2d)
plot(mod2dr)
summary(mod3dr)
plot(mod3dr)
plot(mod3dr)
Using the palette
Most drawings use the default color.
The palette can be defined and assigned to the above options as follows.
options("Colors" = Palette )
Pairs(iris\[,-5\], Species)
If desired, users can easily define their own palettes.
reference
Fraley C. and Raftery A. E. (2002) Model-based clustering, discriminant analysis and density estimation, _Journal of the American Statistical Association_, 97/458, pp. 611-631.
Most popular insights
1.Application case of multiple logistic regression in R language
2.Case implementation of panel smooth transition regression (PSTR) analysisAnalysis case implementation “)
3.Partial least squares regression (PLSR) and principal component regression (PCR) in MATLAB
4.Case study of Poisson Poisson regression model in R language
5.Hosmer lemeshow goodness of fit test in R language regression
6.Implementation of lasso regression, ridge ridge regression and elastic net model in R language
7.Implementing logistic logistic regression in R language
8.Prediction of stock price by linear regression in Python
9.How R language calculates IDI and NRI indexes in survival analysis and Cox regression