For this lab, I am using clustering to determine which NBA players are being underpaid based on 6 performance metrics, which were chosen based on their high correlation (greather than 0.50) with salary. The metrics I chose were:
Games
Field Goals
Free Throws
Assists
Turnovers
Points
Then, I standardized the variables and performed a clustering analysis to determine the ideal number of clusters and the most underpaid but highest-performing players.
This data is from 2020-2021 NBA season and can be found on basketball-reference.com.
The elbow plot shows the explained variation as a function of the number of clusters. In the plot below, there is a sharp increase from 1 to 2 clusters and another smaller, but significant, increase from 2 to 3 clusters. Therefore, I would choose to do a cluster analysis with 3 clusters based on the results from the elbow plot.
Based on this plot, the optimal number of clusters is tied at 2 or 3. Since we concluded with the elbow plot that there was a significant increase between 2 and 3 clusters, I will choose to use 3 clusters for my analysis.
Based on my analysis with 3 clusters, I have selected 5 high-performing players that appear to be underpaid. As evident in the Points vs Assists by Salary graph, Cluster 2 is the top-performing cluster. I created a visualization with only the players in Cluster 2 that have more assists and points better than the median assists and median points, respectively, and are paid below the 75th percentile for salary. In this visualization, titled Highest-Performing Players, I have selected the five players with the smallest salaries, which appear to be great candidates for future draft picks. The players are:
Shai Gilgeous-Alexander
Bam Adebayo
Coby White
Collin Sexton
Donovan Mitchell