When we run our student schedules we only have 45% fulfilled and some students are not not being put into classes. What am I doing wrong?

Rachel Thacker

When we run our student schedules we only have 45% fulfilled and some students are not not being put into classes. What am I doing wrong?

Scott Chrysler

@Rachel Thacker - Rachel, you are not doing anything wrong. You are experiencing some of the limitations of the scheduling module. It has been my experience that when I have tried to let "the system" build the entire schedule from scratch, the output is marginal at best. The schedule maker does not have the "brains" to keep cycling through the data to rearrange and work out conflicts until it reaches an optimum schedule. I have always used a hybrid model of hand scheduling and Blackbaud. Typically, I will hand-schedule the 10 students who have the most challenging course requests, meaning all of the courses they requested are singletons or mostly singletons and doubletons. This is not that difficult to do using the conflict matrix. This establishes the foundation of my master schedule, and then I build out all of the singletons, then doubletons, then tripletons, etc via what I call "chunking." Chunking is taking the next group of 4 or 5 courses with the most requests and having the schedule maker put them in the schedule. I then populate those sections with the students who have requested them. Once the students are placed, I check the fulfillment matrix to ensure all students are placed. I keep chunking until the master schedule is complete. Then I go back to balance section numbers. I should also add that before I do any of this, I have manually created all of the sections for the year and assigned teachers and rooms to those sections. We know going in we will have 3 sections of AP Calc AB or 1 section of Linear Algebra and 3 sections of Geometry Honors. I create all of those sections and assign the teachers and room numbers. If we are not sure who is teaching a section, I use administrators who do not teach as place holders and rooms that are typically not scheduled as locations. Once we know who is teaching or where an available room may be, we make the change.

So again, you are not doing anything wrong. The Blackbaud schedule maker is just not that sophisticated in my experience. It will be interesting to see whether Blackbaud chooses to enhance the scheduling module with AI in the near future. In my opinion, it could use an overhaul.

Good luck.

Scott

Rachel Thacker

Scott, thank you so much. Yes we do most of the work before running in BB. However this year it seems we have a number of students just not being put in classes. I was just thinking I missed a step or something.

Scott Chrysler

Rachel - Check your Course limits numbers. We ran into several cases this spring in which the scheduler did not add students after the target number was reached, even though the Maximum number was 2 or 3 students more. Rather than place the students and hit the maximum, it left them unscheduled, and we needed to add them by hand. That may account for some of the students.

Is there any commonality in the non-placed students or the courses in which they are not being placed?

Scott

Rachel Thacker

Thank you again

Yes, that seems to be what is happening, and no, no commonalities.

Shirley Wagar

@Rachel Thacker 45% fulfillment? Luck you! I couldn't get beyond 30%. I took all of my scheduling data (student choices, teaching assignments, room constraints, etc.), ran it through ChatGPT and created a schedule that allowed me to place 93% of my students into conflict free schedules. It was astounding to find that an AI tool could allow me to create a master timetable that couldn't be done using Blackbaud's own Scheduling software. I'm hoping that Blackbaud embraces the use of AI to redevelop their schedule maker software because, as of right now, it's not a reliable option for scheduling.

Rachel Thacker

Wow, that is crazy. What format did you use for student choices, teaching assignments, room constraints, when you ran it through Chat ?

Shirley Wagar

@Rachel Thacker It was a multi-tabbed Excel Spreadsheet (.xlsx).

The student choices (one row per user_id) were on one tab, another tab had a list of teachers and their assigned course/section, a third tab had available rooms and room types, and a final tab listed constraints (i.e. Mr. Smith can only teach Grade 9 Science in Room 211 during Block 1).

We have an 8-block schedule and offer 97 courses (161 sections). It took ChatGPT a few tries, but after each pass, I would clarify something, or suggest another approach, and it was incredibly responsive. When it finally produced what I felt was a usable schedule, I imported the data and manually managed conflicts. I was truly impressive. I think I completed the entire scheduling process in under a week!

Rachel Thacker

Wow thank you so much we might just try this.