Reference

Methodology & data definitions

Everything in ARIA is computed from one deterministic synthetic dataset of 35,000 fictitious alumni (fixed seed — identical every session).

Definitions

Total alumni: Synthetic alumni graduated on or before the reporting year, after global filters.
Active alumni: Any recorded engagement within the last 12 months.
Employment rate: Employed or founder/self-employed ÷ alumni not retired or in further study.
Executive: Current seniority VP or C-Suite.
Career Advancement Index: Mean of actual seniority ÷ expected seniority for years of experience (100 = par).
Time / Treasure / Talent: Each 0–100, a saturating function (100·(1−e^(−x/k))) of underlying activity counts or giving.
Engagement index: Weighted mean of the three pillars using the configurable weights (default 40/30/30).
Segments: Highly Engaged (≥55, active ≤1 yr) · Consistently (≥32, ≤2 yrs) · Dormant (no activity 5+ yrs) · At Risk (3–4 yrs) · Emerging (≤6 yrs since grad, ≥12) · otherwise Underengaged.
Opportunity priority: 100 × (0.55 × probability + 0.45 × institutional relevance); best category per alumnus.
Compensation: Synthetic benchmark bands by seniority × industry × region — not verified individual earnings.

Predictive models

Logistic: p = 1 / (1 + e^−(β₀ + Σ βᵢ xᵢ)); features scaled 0–1. Coefficients are illustrative.

Executive Advancement Propensity

Likelihood of reaching VP / C-Suite level within the forecast horizon.

β₀ = -4.6 · seniority +4.2 · experience +1.4 · mobility +1.3 · relationship +0.5 · founder +0.4 · events +0.3

Entrepreneurship Propensity

Likelihood of founding or co-founding a venture.

β₀ = -3.4 · founder +3.2 · mobility +0.9 · seniority +0.8 · experience -0.6 · events +0.4 · corporate +0.3

Mentorship Participation Propensity

Likelihood of accepting a mentoring commitment in the next academic year.

β₀ = -3.3 · mentoring +2.6 · frequency +1.6 · seniority +1.4 · relationship +1.1 · outreach +1.2 · recency +0.8 · events +0.6

Philanthropic Participation Propensity

Likelihood of making a gift in the next campaign cycle.

β₀ = -3.6 · giving +3 · seniority +1.4 · relationship +1.2 · experience +0.9 · recency +0.8 · outreach +0.9 · events +0.4

Alumni Disengagement Risk

Likelihood of no meaningful engagement over the next 24 months.

β₀ = 1.1 · recency -2.4 · frequency -1.8 · relationship -1.5 · events -0.8 · outreach -1.1 · mentoring -0.6 · mobility +0.3

Corporate Partnership Opportunity

Likelihood of facilitating a corporate partnership (recruiting, sponsorship, project).

β₀ = -4.4 · seniority +2.8 · corporate +2.2 · relationship +1 · outreach +0.9 · frequency +0.7 · founder +0.4

Explanations use βᵢ·(xᵢ − x̄ᵢ), the exact additive attribution for a linear logit. Performance metrics (AUC, precision, recall, Brier, calibration) are computed against seeded synthetic outcomes to demonstrate the evaluation pipeline. All relationships are associations — not causal effects.

Architecture

Data generation, analytical calculations, scoring models and UI are separate modules. Activation workflow changes persist in the browser; the data layer is structured so a backend can replace the generator later.