######################################################
# Script to create dendrograms and clusers in R

# read the data into a datatable
TheData <- read.csv('C:/Temp/599_LabClusters/GEO499_580_cluster_analysis_data_20130225_9PM.csv')

#this is new, Doug
XY <- c(TheData$Current_Residence_Long,TheData$Current_Residence_Lat)
XY <- c(TheData$Hometown_Long,TheData$Hometown_Lat)

Matrix <- matrix(XY,23,2)

######################################################
# Run this section to see the points and the center of the clusters

# Prepare Data
TheData <- na.omit(Matrix) # listwise deletion of missing
Matrix <- scale(Matrix) # standardize variables

# K-Means Cluster Analysis
ClusterInfo <- kmeans(Matrix, 4) # matrix of x and y values, number of clusters

# Plut the result using the cluster to colorize the points in each cluster
plot(Matrix, col = ClusterInfo$cluster)

# add the centers of the clusters
points(ClusterInfo$centers, col = 1:4, pch = 8, cex = 2)

######################################################
# Run from here to get the dendrogram (elbow)
# this is stocastic so run it repeatedly to see the variation

# find the varience 
TheVariance=apply(Matrix,2,var)

# find the sum of squares for 1 cluster
WithinClusterSumOfSquares <- (nrow(Matrix)-1)*sum(TheVariance)

# find the sum of squares for 2 to 15 clusters
for (i in 2:15) {
  ClusterInfo=kmeans(Matrix, centers=i)
  WithinClusterSumOfSquares[i] <- sum(ClusterInfo$withinss)
}
# plut the result
plot(1:15, WithinClusterSumOfSquares, type="b", xlab="Number of Clusters",ylab="Within groups sum of squares")