Practical Predictive Analytics

Running k-means to generate the clusters

Run kmeans function using the kcca package to generate five cluster assignments:

# repeat kmeans using the kcca function. Clusters=5 clust1 = kcca(dtMatrix, k = 5, kccaFamily("kmeans")) clust1 > kcca object of family 'kmeans'  >  > call: > kcca(x = dtMatrix, k = 5, family = kccaFamily("kmeans")) >  > cluster sizes: >  > 1 2 3 4 5  > 360 120 152 387 8981

Print the number categorized for each cluster:

table(clust1@cluster) >  > 1 2 3 4 5  > 360 120 152 387 8981

Merge the clusters with the training data, and show some sample records displaying the cluster assigned to each:

 kw_with_cluster2 <- as.data.frame(cbind(OnlineRetail, Cluster = clust1@cluster)) head(kw_with_cluster2) > InvoiceNo StockCode ...