Entry 10Presenting the Relationship between Students’ Interests and Selected Majors using A Matrix-based Visualization
Xiaoyao Xu, Jia-Kai Chou, and Kwan-Liu Ma — University of California, Davis

We use a matrix-based visualization to present the relationship between student se- lected majors and students’ best-fit majors. Note that, a gamma expansion (γ = 2.5) was applied to the ratios of majors to keep the cell size of the least popular major (“A&M”) noticeable. The best-fit ratio measures how well the students’ selected ma- jors match their true interests. A bigger value on a diagonal cell of the matrix means more percentage of students in that particular major finds their best-fit. In addition, the order of the majors is sorted in descending order according to the best-fit ratios. The curved lines between matrices represent students’ change of majors, while dark gray lines indicate larger percentage of major changes (> 10%). On the top of each matrix, we use bar charts to compare other attributes. From the visualization, we find:

  1. The most popular majors (bigger cell size) are not shown to be the most best- fitting majors (darker cell color).
  2. “BUS” has the highest percentage of best-fit students over the three years.
  3. “BUS”, “EDU”, and “C&P” are the top three best-fitting majors.
  4. A rather high percentage of students majors in“Sci” but actually best-fits “H&T”.
  5. “P&T” and “ARC” students usually cannot find their best interest well.
  6. From high school to first year college, “H&A” and “E&D” face the most sig- nificant overall outflux in terms of percentage, while the most significant single outflux of students changes from “H&T” to “SCI”.
  7. If a student’s best interest is in “ENG”, then he/she is very likely to choose “ENG” as major.
  8. Most majors have more students from 4-year college than 2-year college except “S&L”.
  9. There is higher percentage of “SCI” and “ART” students that finds their good-fit comparing to other majors.
  10. The most female/male-preferred majors can also be easily identified.

* There are several other findings can be found in our visualization, we leave them out because of the word limit.

Xiaoyao Xu (sphxu@ucdavis.edu)
Jia-Kai Chou (jkchou@ucdavis.edu)
Kwan-Liu Ma (ma@cs.ucdavis.edu)
University of California, Davis
1 Shields Ave, Davis, CA 95616