UCSB Passtimes Pilot

Transcript PDF Upload
Pass 1 and Pass 2 Recommendations
Pass 3 Recommendations
Time-series Enrollment Data For CMPSC 16
Dark mode

For Gauchos, course registration is an endless dilemma of having to blindly choose a few classes during pass 1, choose the remaining classes in pass 2 and 3, and pray that there's still a spot left when your pass comes.

Having continuously monitored enrollment trends data throughout the UCSB 2024 Winter Registration, we are proud to announce that having blindly signing up for classes is finally a bygone era.

Introducing the Gaucho Passtimes Pilot

Using time-series data of how quickly each class filled up during Winter Registration, we created the Gaucho Passtimes Pilot (GPP), a game-changing web platform that helps you select the right class at the right passtime.

GPP houses a collection of 10k+ course prerequisites and major requirements dataset to power its recommendations.

To unlock the full potential of GPP, simply upload your transcript as a PDF to the front page and apply corrections as needed.

Having known which classes you’ve taken, GPP then generates a full list of recommended classes you should take in the coming quarter using topological ordering, centrality, and natural language processing algorithms. GPP intelligently prioritizes these classes based on how fast this class filled up in the past quarter, how many future classes it will satisfy the prerequisites for, and many more weighting criteria.

To visualize how fast each class filled up, GPP allows you to browse the enrollment time-series data in a beautifully presented diagram.

You may also directly search for the enrollment trends of any class under the “enrollments” page with autocomplete suggestions.

And, everything works like a charm in dark mode 🙂

Challenges Faced and Overcame

Our data source – UCSB SA Catalog – was highly irregular. For example, on the Catalog, prerequisites for a course may be listed as CMPSC 16 and MATH 4B, which is relatively straightforward. But for other classes and majors, the prerequisite texts may be just 16 and 4B without any prefix information or 16 and 4B or 40 and 41, 42 of heavily nested boolean relations.

Beyond that, a professor’s name could appear as a full name on one page but a completely different nickname on another; sometimes there’s one professor only, sometimes there are multiple. Sometimes the professor's information is straight-up missing. Sometimes different professors’ names are the same, even within the same department.

To reconcile the irregular prerequisites information, we resorted to the help of natural language processing and simple regex matchings. Still, there remained instances of irresolvable confusion, in which case we manually annotated these data. For professor names, we relied on SQL queries and probability to determine if “Joe B” more likely refers to “Joe Bob” or “Joe Bennet” given all course names and their correlations with professor names.

What's Next?

Our next ambition is to:

Create a one stop shop where Gauchos can not only view enrollment trends and recommend passtimes, but also to read and create professor reviews, see grading trends for each class, and create class schedules with AI Assistant.

Our ultimate dream is to: