You are using k-means clustering to discover groupings within a data set. You plot within-sum-of-squares (wss) of multiple cluster sizes. Based on the exhibit, how many clusters should you use in your analysis?