// module overview
The dataset contains 123 "lunch-only attendance" events — records where a student was
marked present for the lunch period but absent for all academic periods. When you plot
these events by day of month, the pattern is unmistakable: a spike after day 20.
That spike is the SNAP benefit exhaustion window. Families often run out of food assistance in the last week of the month. Some students come to school for the meal and leave. The data calls this an "attendance issue." It isn't.
This module teaches you to find that pattern — and think carefully about what to do with it. It's one of the most ethically important modules in the curriculum.
That spike is the SNAP benefit exhaustion window. Families often run out of food assistance in the last week of the month. Some students come to school for the meal and leave. The data calls this an "attendance issue." It isn't.
This module teaches you to find that pattern — and think carefully about what to do with it. It's one of the most ethically important modules in the curriculum.
// key insight
A kid who shows up for lunch but not for class isn't truant — they're hungry. The data has been telling us this the whole time.
// what you'll learn
What Educators Will Learn
- How food insecurity manifests in school attendance and engagement data
- The SNAP benefit cycle and why the end of the month matters for some families
- Why labeling this as an 'attendance problem' causes harm — and what to call it instead
- How to use this information to connect families with support resources, not consequences
- The ethics of identifying vulnerable students through data: privacy, dignity, and confidentiality
Python Walkthrough
- Isolating lunch-only attendance events from the full attendance table
- Extracting the day-of-month and plotting a histogram — watching the spike emerge
- Correlating lunch-only events with food insecurity flags in the student context table
- Measuring the statistical significance of the end-of-month clustering
- Building a privacy-safe summary report: pattern-level data without individual student exposure