Using the data from College Basketball Analysis Powered by SAP Lumira & Predictive Analytics I used SAP Predictive Analytics 2.0 with Expert Analytics.
The above shows the Designer view from the Predict tab. The configuration was very easy; input is the data from the link above, the algorithm is automatic clustering, and the output of results is to a CSV file (why? I will explain later). You learn more about clustering at this Wikipedia page: Cluster analysis – Wikipedia, the free encyclopedia
After executing the model, I am prompted to switch to the results view.
Scrolling to the right, I see the Cluster column has been added.
Above is the algorithm summary, with 10 clusters. Interesting to me is that it only took 1 second of “learning” time.
Above is the cluster distribution.
The Parallel Coordinates chart is shown above. Only Kentucky has 0 losses, so they can be easily traced.
The scatter matrix chart is displayed above.
I create a slide show story but when I go to share, I cannot Share to Lumira Cloud as shown above.
So I take the CSV output of the clustering and take it to Lumira 1.23
Of course, I want to know where my team, Kansas, is in the clusters – cluster 4 as shown above.
Kentucky, along with great teams like Duke, Virginia, and Notre Dame is in cluster 8. I would expect the teams in cluster 8 to be selected today.
Because I am using Lumira I can publish to the Lumira Cloud, and the story is here.
The data does not take into account the conference tournaments that happened this weekend.
The above is from this past week’s BI 2015 conference, which was the location also for the PAC 12 Basketball Tournament. The above shows the Stanford Pep Band, along with the Stanford tree mascot.
Join ASUG this Thursday for the What’s New in Predictive Analytics 2.0 webcast – register here.