CSU Cluster Analysis: Geocoded Data in Tableau

Michael Le
Research Associate
Humboldt State University

Educating 474,600 students in 2016, the California State University (CSU) system is the nation’s largest four-year public university system with 23 campuses. Humboldt State University (HSU) is the Northern most campus. With a distance of 233 miles to the nearest sister campus, HSU is the most remote campus. Over 90% of new students in 2016 came from outside the local service area. HSU’s incoming class came from over 2,000 different California high schools and community colleges. With limited resources, HSU’s Enrollment Management needed a strategic way of selecting sites to host recruitment events.

A visual inspection of HSU’s 2,033 California feeder high schools and community colleges showed four distinct groupings (Sacramento, San Francisco, Los Angeles, and San Diego).


We used Tableau to generate longitude and latitude coordinates for each school site (based on ZIP Code). This data was then exported to Microsoft Excel and longitude and latitude were converted into decimal numbers (cluster analysis does not work on fields that are geocoded). The new data set was imported back into Tableau and a Cluster Analysis was performed.

Several models were examined before choosing a model with 18 clusters. Six of the clusters included a significant number of schools (n= 129 to n=435). The six clusters were coded with a unique dark color and all other clusters were coded a light blue. The six clusters of interest were Sacramento (C8), San Francisco (C11), Los Angeles (C14), Orange County (C16), Inland Empire (C13), and San Diego (C18). While the visual analysis was fine, the amount of time and effort it took to create a cluster analysis and examine several models was minimal. Tableau’s Cluster Analysis feature shows that it behooves Tableau users to conduct such an analysis if appropriate for their data.



To add a comment, Sign In
There are no comments.