* do-file for Additional multivariate exercise 4, VHM 802, Winter 2021
version 16 /* works also with versions 14-15 */
set more off
cd "r:\

* Manly Example 1.4
import delimited r:\prehist_dog.csv, clear
* create grouping variable with shorter names
gen group0=group
replace group0="M.dog" if _n==1
replace group0="Jackal" if _n==2
replace group0="C.wolf" if _n==3
replace group0="I.wolf" if _n==4
replace group0="P.dog" if _n==7
* standardize variables
foreach var of varlist mandbreadth-mol14length {
  egen `var's=std(`var')
  }

* hierarchical cluster analysis
cluster single mandbreadths-mol14lengths, name(singledogs) measure(L2) /* L2 is the default for single linkage */
cluster dendrogram singledogs, label(group0) title("Prehistoric dogs: Single linkage") ytitle("L2 dissimilarity")
cluster average mandbreadths-mol14lengths, name(averagedogs) measure(L2) /* L2 is the default for average linkage */
cluster dendrogram averagedogs, label(group0) title("Prehistoric dogs: Average linkage") ytitle("L2 dissimilarity")
cluster complete mandbreadths-mol14lengths, name(completedogs) measure(L2) /* L2 is the default for complete linkage */
cluster dendrogram completedogs, label(group0) title("Prehistoric dogs: Complete linkage") ytitle("L2 dissimilarity")

* K-means clustering
set seed 210315
forval i=1(1)10 {
  cluster kmeans mandbreadths-mol14lengths, k(2) start(krandom) name(dogs2m`i')
  tab group dogs2m`i'
  scalar wss=0
  foreach var of varlist mandbreadths-mol14lengths {
    quietly anova `var' dogs2m`i'
	scalar wss=wss+e(rss)
    }
  di "Within-cluster sum-of-squares: " wss
}
forval i=1(1)10 {
  cluster kmeans mandbreadths-mol14lengths, k(3) start(krandom) name(dogs3m`i')
  tab group dogs3m`i'
  scalar wss=0
  foreach var of varlist mandbreadths-mol14lengths {
    quietly anova `var' dogs3m`i'
	scalar wss=wss+e(rss)
    }
  di "Within-cluster sum-of-squares: " wss
}