get_tree_fit {lazy.cluster} | R Documentation |
Calculate RSS etc for each # of clusters
get_tree_fit(data = NULL, tree = NULL, print = 1, plot = 0)
data |
data frame |
tree |
A class hclust object created by dhclust |
print |
= 1 to print the result |
plot |
= 1 to plot the tree with rss as the height |
Given k clusters, k=1,2,...,nrow(data), the data matrix Y
can be expressed as
Y = X %*% B + E
where X is the nrow(Y) times k design matrix consisting of 1s and 0s,
and B is the k times ncol(Y).
RSS associated with this model, namely,
RSS = tr( t(Y)%*%(I-P(X))%*%Y )
, where P(X)=X%*%inv(t(X)%*%)%*%t(X)
will be returned.
Assuming multivariate normality of the rows of the E matrix,
minus twice the log likelihood will be returned as m2llh.
A class hclust object with rss, m2llh, aic, and bic added.
# native hclust reshc <- hclust(dist(unidimdata$y)) temp <- get_tree_fit(unidimdata$y,reshc, plot=1) # divisive resdhc2 <- dhclust( unidimdata, ndiv=2, plot=1, print=1, method="LS" ) temp=get_tree_fit( unidimdata, resdhc2, print=2, plot=1 ) resdhc32 <- dhclust( unidimdata, ndiv=3, plot=1, print=1, method="LS" , maxclustersize=2 ) temp=get_tree_fit( unidimdata, resdhc32$tree_s, print=2, plot=1 )