// module overview
Most schools wait for report cards to identify students with academic struggles.
By then, a student may have missed 6–8 weeks of instruction and fallen significantly
behind. The dataset contains assignment-level grade records — individual scores,
not just final averages. That granularity is the key.
In this module, you'll build a rolling GPA trend from individual assignments, identify the specific week a student's trajectory started to decline, and compare that early signal to when a traditional grade-threshold flag would have fired. The gap is often 3–5 weeks — and in a school setting, that's the difference between a short conversation and a formal intervention.
In this module, you'll build a rolling GPA trend from individual assignments, identify the specific week a student's trajectory started to decline, and compare that early signal to when a traditional grade-threshold flag would have fired. The gap is often 3–5 weeks — and in a school setting, that's the difference between a short conversation and a formal intervention.
// key insight
The week a student's grade trajectory flips negative is almost always 3–5 weeks before anyone gets a formal notification. That gap is your intervention window.
// what you'll learn
What Educators Will Learn
- Why report card GPA is a lagging indicator — and what a leading indicator looks like
- How assignment completion rates predict grade trends before they materialize
- The difference between a student having a bad week and a student in a declining pattern
- How to use trend data in teacher-advisor conversations without alarming students or families prematurely
- What 'regression to the mean' is and why it can fool you into thinking an intervention worked
Python Walkthrough
- Aggregating assignment-level scores into a weekly GPA proxy using weighted averages
- Building a 4-week rolling GPA trend per student per course
- Detecting inflection points: the week the slope turned negative using
np.gradient() - Comparing early warning trigger date (trajectory) vs. late warning (threshold breach)
- Creating a multi-student comparison chart: who is declining, by how much, and since when