Methodology
How ACTI Analytics classifies activity and calculates distraction metrics.
How distraction is classified
The ACTI agent runs on each student device and captures the foreground app or website every five seconds. Each activity segment is assigned a category based on the app or domain.
Each segment is classified using an AI classification system. A segment is educational if its category falls within a defined set of productive categories. Everything else is classified as distracted. For example, Google Docs is educational. YouTube or Instagram is distracted.
Educational categories
Student distraction %
The percentage of the school day a student spent on non-educational apps and websites.
Total secondsis the school's configured hours, typically something like 8:30 am to 3:30 pm. Activity before or after these hours is not counted (or recorded). Segments that span the boundary are clipped to the window.
Example
A school day runs from 8:30 am to 3:30 pm. That's 7 hours, or 25,200 seconds. If a student spent 15,120 seconds on Instagram, YouTube, and gaming sites:
That student's distraction for the day is 60%.
The school-wide average is calculated across all students who had device activity that day. Students with no recorded activity are excluded from the average.
Teacher distraction %
The average distraction across all enrolled students during a teacher's scheduled periods. This measures student behaviour during a teacher's classes, not the teacher's own device usage.
Supervised student-minutes is the period duration multiplied by the number of enrolled students.
Distracted student-minutesare calculated by adding up the distracted minutes from each individual student in the class to give an aggregate measure of distraction in that teacher's classroom.
Example
A teacher has a 50-minute period with 25 students.
During that period, each student's distracted minutes are added together. Say the total across all 25 students is 375 distracted student-minutes:
That teacher's distraction for the period is 30%.
Only teaching periods are included. Recess and lunch are excluded. Segments are clipped to period boundaries so activity outside a period does not affect that teacher's metric.
Because this metric sums distracted minutes across all students, it can be skewed by a small number of highly distracted outliers. To account for this, we also show an IQR distractionfigure — this removes statistical outliers (values beyond 1.5× the interquartile range) before calculating the average, giving a more representative picture of typical classroom behaviour.