An elbow point marks where adding clusters produces sharply diminishing reductions in within-cluster variance. A scree plot places cluster count on the x-axis and within-cluster sum of squares on the y-axis; the bend suggests a practical value for K. section 11.5.2
Treat the elbow as a judgment rather than a guaranteed answer: scree plots may lack a clear bend, and diminishing returns may not identify the same K. Compare the result with the average silhouette method, which uses coefficients from -1 to +1 to assess cluster separation. section 11.5.2
NBA payrolls and wins
Using factoextra in R, compute within-cluster-sum-of-squares and silhouette plots with fviz_nbclust(), then compare them vertically. A subsequent k-means solution uses K = 4 to group standardized team payrolls and regular-season wins. section 11.5.2
