CMSC 838S/828S: Information Visualization

Hench Qian (qian@cs.umd.edu)
Application Project:
Visualizing the University Course Listings
October 5, 1999

Introduction | Data Set | Searching and Browsing | Other Visualizations | Critique

Introduction

The university course listing is arguably the most important publication published by the university. Available online or on paper, it is the centerpiece of every student's schedule, around which every other activity revolves. It details every class meeting in every subject by every instructor in every room in the university. Yet, despite its importance, the course listings are presented in a remarkably clumsy fashion. Students are forced to look at hundreds of conflicting course listings over and over before a satisfactory course schedule can be assembled. Furthermore, the wealth of information contained in the listing lie largely unexamined. In this project, the Spotfire visualization tool is used to search and browse the course listings, and explore and visualize the wide array of information available in the listings.

Data Set

The data for this project is the 1999 Fall semester course listings available online. The listings are parsed to extract the following information:

Courses:
- Department
- Course Number
- Title
- Credits
- Sections		 Sections:
- Section Number
- Course Code
- Lecture (Time and Location)
- Discussion (Time and Location), if any
- Seating		 Meetings:
- Day of Week
- Start Time
- End Time
- Building and Room
- Meeting Type (Lec, Dis, Lab)

This information is kept for each course, section, and class meeting. The complete data set is available here. The number of entries are given below:

 

University of Maryland

CMPS

Courses

3502

294

Sections

5543

735

Meetings

13577

2324

Because of the large number of classes university-wide, this project is focused on courses in the College of Computer, Mathematics, and Physical Sciences (CMPS) in order to make the project more manageable.

Searching and Browsing

The primary use of the course listings is for course scheduling. In order to assemble a desirable course schedule, students would like to browse and search the course listings based on a number of attributes and filter irrelevant or conflicting listings. Current presentations only allow students to browse by department and course number. Spotfire can be used to allow searching and browsing based on any combination of department, course number, title, instructor, day / time, location, and seating.

Example: List all sections of 400 level and higher CMSC courses with open seats.

Step 1: Overview

Initially, Spotfire provides an overview of all of the sections available. It displays every section by its location on the map. Size, color, shape, and rotation can be used to represent different features of the section. Jittering is applied to somewhat overcome occlusion.
Step 2: Filter

Spotfire dynamically filters out listings as the user changes query values. First, deselect all departments except CMS. Next, set the lower bound on course numbers to 400. Finally, set the lower bound on open seats to 2. At this point, only the items satisfying the query criteria remain.
Step 3: Zoom

After selecting the remaining items, Spotfire provides a detailed listing of the selected items in text or HTML (Here, we have HTML).

Similarly, Spotfire can be used to instantaneously answer other common queries that are helpful in making scheduling decisions:

Other Visualizations

Aside from being used to assemble schedules, the course listings contain a wealth of information that largely goes unexplored. Spotfire presents a previously unavailable tool for exploring this information. Here are some interesting observations that can be made:

1. Course Offerings by Department.

Which department teaches the most classes? A histogram shows that the Math department offers by far the most sections. However, by adjusting course numbers, it can be shown that the number of course offerings even out as the course level increases.
2. Course Level vs Students

Is it true that lower level courses have more students than upper level courses? Yes, a histogram overwhelmingly shows that lower level courses have many more students than upper level courses.
3. Class Occupancy

What are the most and least popular courses in CMSC? A scatter plot of course vs class occupancy (%) shows that most classes are filled to capacity, but there are several classes in low demand (760, 818Z, 251).
4. Department vs Class Location

Where are all the classes for each department concentrated? A scatter plot on a campus map shows the concentration of classes. Most courses by a department are concentrated in a few buildings, with the rest scattered here and there.

These are just some of the observations that can be made. Spotfire provides almost an infinite number of combination to visualize.

Critique

Spotfire is a very useful tool for visualizing the course listings. It is much easier and efficient to use than existing tools (paper and online). It also has some obvious advantages against other visualization tools (Excel, database queries). Here is a comparison of some of Spotfire's features and problems against the alternatives:

Spotfire

Alternatives
Positives:
Dynamic Querying - The instant response provided by Spotfire is its most important feature. Not only does it increase productivity, it also encourages exploration. The alternatives require scanning through the entire listings or entering in a new database query for each different query.
Screen Usage - Spotfire can provide a comprehensible display of hundreds to thousands of items on the screen at once using position, color, size, shape, and rotation to represent attributes. The alternatives can only show on the order of tens of listings in a single browser window or table listing.
Negatives:
Large Data Sets - Spotfire loses some effectiveness with large data sets. The control panel gets large, and it takes time to scroll through the query devices. The legends become cluttered. Also, the sliders become more difficult to use as higher precision is required. The alternatives become more difficult to use as well as the data sets increase in size.
Occlusion - When many items reside at the same point on the screen, occlusion becomes a problem. While jittering partially solves the problem, it is not a total fix. Tabular approaches do not have this problem, but other graphical approaches have similar problems as well.
Representing Ranges - Spotfire cannot represent a range (e.g., a class period). Some tinkering must be done (e.g., connecting lines) to achieve the effect of a range. Tabular approaches do not have this problem, but other graphical approaches have similar problems as well.
Discrete (Unordered) Data - Spotfire automatically sorts discrete data alphabetically when it may not be desired (e.g. weekdays). It is sometimes desired to be able to specify an order so that sliders would work. Other approaches do not have this problem.

Spotfire provides very good mechanisms to efficiently search and browse for courses and allows the further exploration of information contained within the listings. Its positives far out-weigh the negatives. In all, Spotfire is a very effective tool for visualizing course listings.

Web Accessibility