Visualization of Enrollment data using Chord Diagrams - GRAPP 2015
2 likes1,679 views
This document discusses visualizing enrollment data from an open university using chord diagrams. It analyzes subject combinations and visualizes intra-semester and inter-semester enrollment patterns. Chord diagrams show relationships between subjects as segments connected by chords. This allows exploration of popular subject combinations and pass/fail rates. The visualization is interactive, providing subject and combination details. While effective, it has limitations like complexity with many subjects and lack of data for uncommon choices. Future work includes evaluation, alternative visualizations, and integrating it into the enrollment process.
![Chord parametrization
● Number of segments
– Number of chords grows quadratically
– Up to 10 segments looks fine, 6-8 seems optimal
● Segment colors
– From red (FAIL) to green (PASS) through yellow
– Other thresholds can be used: [0, 0.4, 0.6, 1]
● Chord colors
– Percentage of learners passing the two subjects
– Same red-yellow-green coloring
Visualization of Enrollment Data using Chord Diagrams – GRAPP 2015, Berlin, Germany](https://image.slidesharecdn.com/grapp2015-92-150314025217-conversion-gate01/85/Visualization-of-Enrollment-data-using-Chord-Diagrams-GRAPP-2015-10-320.jpg)
![Thank you!
jminguillona[at]uoc[dot]edu
twitter/@jminguillona
http://bit.ly/1wcZCBk
Visualization of Enrollment Data using Chord Diagrams – GRAPP 2015, Berlin, Germany](https://image.slidesharecdn.com/grapp2015-92-150314025217-conversion-gate01/85/Visualization-of-Enrollment-data-using-Chord-Diagrams-GRAPP-2015-15-320.jpg)
Visualization of Enrollment data using Chord Diagrams - GRAPP 2015
- 1. Visualization of Enrollment Data using Chord Diagrams L. Blasco-Soplon, J. Grau-Valldosera, J. Minguillón Universitat Oberta de Catalunya Barcelona, Spain
- 2. Context ● UOC: fully online open university – Learners with very diverse backgrounds – No enrollment requirements ● How many subjects? ● Which ones? – Recommendations from ● Subject / course / degree planning (text / table based) ● Mentor ● “Common sense” ● First enrollment is known to be “critical” Visualization of Enrollment Data using Chord Diagrams – GRAPP 2015, Berlin, Germany
- 3. Learning Analytics ● Plenty of available data – Since 1994 (EHEA / Bologna degrees since 2008) – Thousands of students each semester ● Hundreds per degree – Tens of thousands of subject combinations – Academic performance ● Subject level (PASS / FAIL) ● Semester level (dropout) ● Can we do it better? Visualization of Enrollment Data using Chord Diagrams – GRAPP 2015, Berlin, Germany
- 4. Our proposal ● Analyze subject combinations – 2 x 2 contingency tables (PASS / FAIL) ● Visualize enrollment data – Intra-semester – Inter-semester ● Provide better support (esp. 1st semester) – Learners: adjust enrollment – Mentors: improve recommendations – Degree managers: reduce dropout Visualization of Enrollment Data using Chord Diagrams – GRAPP 2015, Berlin, Germany
- 5. Available data (I) ● Paper: Business and Economics degree – From Spring 1999 to Spring 2011 (25 semesters) – 21792 learners – 501 different subjects – 328467 subject enrollments ● New data from EHEA B&E degree (10 sems.) – 5930 learners – 64 different subjects (59 in the 1st semester) – 63460 subject enrollments (17610 in the 1st sem.) Visualization of Enrollment Data using Chord Diagrams – GRAPP 2015, Berlin, Germany
- 6. Available data (II) ● First semester only: – High dropout rates – Are learners following institutional advice? ● 5698 learners take 2 or more subjects – 17378 subject enrollments – 59 different subjects ● 13 subjects: 90.6% of learners, 20 subjects: 95.5% – 764 different subject combinations ● 129 combinations: 90.0% of learners Visualization of Enrollment Data using Chord Diagrams – GRAPP 2015, Berlin, Germany
- 7. Chord diagrams ● Circular layout for representing relationships in matrix data as a graph – Segments: nodes – Chords: edges ● Position, color, size ● Popularized in 2007 by the NY Times infographic Close-Ups of the Genome by Giovanni Gherdovich Visualization of Enrollment Data using Chord Diagrams – GRAPP 2015, Berlin, Germany
- 9. Why chords? ● Explore new ways to provide support ● Subjects can be sorted clockwise – From most to least popular subject (size) – Each subject can be independently colored according to its own PASS / FAIL ratio ● Subject combinations – Popularity + % of learners passing / failing one / two subjects ● Available in D3.js – Interactivity “included” as part of the visualization Visualization of Enrollment Data using Chord Diagrams – GRAPP 2015, Berlin, Germany
- 10. Chord parametrization ● Number of segments – Number of chords grows quadratically – Up to 10 segments looks fine, 6-8 seems optimal ● Segment colors – From red (FAIL) to green (PASS) through yellow – Other thresholds can be used: [0, 0.4, 0.6, 1] ● Chord colors – Percentage of learners passing the two subjects – Same red-yellow-green coloring Visualization of Enrollment Data using Chord Diagrams – GRAPP 2015, Berlin, Germany
- 11. Interactivity ● By default, show – Subjects ● Size: Importance (number of learners) ● Color: Perfomance (PASS / FAIL ratio) – Chords ● Size: Number of learners taking such combination ● Color: Number of learners passing the two subjects ● Interaction – On a segment: info about that subject and its combinations with other subjects – On a chord: info about that combination Visualization of Enrollment Data using Chord Diagrams – GRAPP 2015, Berlin, Germany
- 12. Results demo Visualization of Enrollment Data using Chord Diagrams – GRAPP 2015, Berlin, Germany
- 13. Conclusions ● Pros: – Compact visualization, nice aesthetics ● Easy to understand (green: OK, red: KO) – Interactivity reduces data overload – Easy integration into any web based system ● Cons: – 13 subjects generate 13*12/2=78 chords! ● Lack of data for non-common subjects / combinations – Even the simplest visualization needs explanation – What about blind / colorblind learners? Visualization of Enrollment Data using Chord Diagrams – GRAPP 2015, Berlin, Germany
- 14. Current and future work ● Evaluate the visualization compared to tables ● Explore other visualization types – Concentric chord diagrams (inter-semester) – D3.js: parallel sets, ... ● Include other parameters – Learners' satisfaction – Total time spent in the subject – Weight as a inter-semester dropout factor ● Integrate it into the UOC enrollment process Visualization of Enrollment Data using Chord Diagrams – GRAPP 2015, Berlin, Germany
- 15. Thank you! jminguillona[at]uoc[dot]edu twitter/@jminguillona http://bit.ly/1wcZCBk Visualization of Enrollment Data using Chord Diagrams – GRAPP 2015, Berlin, Germany