课弈An international school's G9 options: same 19 sections, largest class from 33 down to 26
We re-ran sectioning on the school's real choice sheet (89 students, 4 blocks): every student kept every choice; with the same 19 sections, physics went from 32/18/16 to 22/22/22 and the class-size standard deviation dropped from 6.0 to 2.3, in about 30 seconds.
Background
The school's Grade 9 has 89 students taking IGCSE options across 4 blocks. Each subject has one teacher, so a subject can run at most once per block, and each block can hold 5–6 sections. The school's manual sectioning already satisfied every choice, but class sizes were uneven: physics ran at 32/18/16.
What we did
We imported the real choice sheet into Keyi with the school's own constraints and solved for three different goals.
| Option | Sections | Largest class | Max same-subject gap | Size std. dev. |
|---|---|---|---|---|
| School (manual) | 19 | 33 | 16 | 6.0 |
| Keyi: same 19 sections, cap 26 | 19 | 26 | 3 | 2.3 |
| Keyi: fewest sections | 18 | 33 | 5 | 5.1 |
| Keyi: small classes, cap 24 | 20 | 24 | 2 | 3.5 |
Results
- Every option kept 100% of student choices.
- With the same 19 sections the largest class fell from 33 to 26; physics became 22/22/22 and chemistry 24/22/22.
- Alternatively the school can save a section, or add one for 24-student classes — its call, with numbers on the table.
- Tiny classes such as Sociology (3) and Drama (5) are set by demand; whether to run them stays a school decision.
What it means
Sectioning goes from weeks of manual work to “set a goal, get an answer in half a minute” — with every trade-off quantified.