SLATE / MEASURES THAT MEAN SOMETHING
Scoring, Classification, and Outcome Prediction Engine
Scores, personas, and predictive estimates that bring student behavior, context, and progress into focus.
Designed and built by David Dysart.
DIFFERENT QUESTIONS / DIFFERENT MEASURES
A student can be highly engaged, academically prepared, and nearly finished with a checklist—and still need help taking the next step.
A school can generate substantial inquiry volume while sending few students to the application stage. SCOPE makes those distinctions visible.
THE CORE PACKAGE
A weighted measure of student interactions, with a defined scale and meaning.
Volume, engagement, and funnel conversion summarized by school, origin source, geography, distance, and other relevant factors.
Estimated likelihood to enroll, using information available at the time of admission.
Interpretable student personas and contextual groupings, including school personas and feeder flags.
SCORING / CLASSIFICATION / OUTCOME PREDICTION
The framework can extend beyond the core package. Additional scores and models are selected around your questions and available data.
SCORING
Keep the scale, inputs, and interpretation clear. An engagement score of 80 does not mean an 80% chance of enrolling.
CLASSIFICATION
Transparent rules explain how a classification was assigned and what it helps the team understand.
OUTCOME PREDICTION
Each model defines its population, outcome, prediction moment, and time horizon.
TWO MODEL PATTERNS / ONE OPERATIONAL HOME
Configurable Joins combines prepared inputs into scores and applies trained model logic to records.
Apply supplied coefficients to the matching inputs and calculate an estimated probability for a defined outcome.
Reproduce supplied splits as conditional branches, following each record to its terminal output.
Model training and validation happen before deployment. In Slate, test calculations and branch paths, then use query exports and scheduled imports to make outputs available for outreach and reporting.
PUT IT TO WORK
Start with the core package or explore a scoring and modeling question.