I was privileged to engage an excellent mathematician/programmer, Jared O. Lovin, for undergraduate research. Jared and I minimized admissions counselors’ travels and used the K-Means algorithm. His work was entitled “Optimization of LaGrange College Admissions Counselors’ Paths Using Student Data“.
I have attached a link to his poster: undergraduateResearchPoster_jaredLovin_final .
Further, I’ve attached a link to his paper that was submitted to LaGrange College’s own undergraduate research journal, Citations. The link to his paper is available at http://www.lagrange.edu/resources/pdf/citations/2015/14_Lovin_Math.pdf .
Although all of our algorithms are sound, we ultimately created an algorithm/process that takes the admissions counselor to each student as his/her physical address (converted to (lat,long) coordinate pairs and then made noisy). Here are some improvements to be made:
- As an interim step, someone needs to create sub-clusters (another K-means sub-implementation) and then run the “Traveling Salesman” algorithm between each of the centroids of these sub-clusters. This creates a more realistic description.
- Interfacing with the Google Maps API is actually fairly easily done. Actual pathways using the info from Maps could truly reduce time and distance.
If anyone is interesting in continuing this research, please reach out to me.