Methodology · Validation · Audit

Every prediction tracked.
Every outcome recorded.
ForVue tells you how accurate it is.

No other predictive maintenance platform closes the loop between prediction and outcome. ForVue does — automatically, continuously, and transparently. The current ±2-4% margin of error is a modeled tier — empirical precision and recall are surfaced in the operator's own accuracy dashboard once 5+ outcome events accumulate for a portfolio.

The Numbers

±2-4%

Modeled Margin of Error

Per confidence tier: HIGH ±2%, MEDIUM ±3%, LOW ±4%. Empirical precision and recall surfaced in the operator's own accuracy dashboard once 5+ outcome events accumulate for a portfolio.

Multi-Factor Scoring

A proprietary set of input factors feed the composite failure probability for every component. Specific factor identities, weights, and thresholds are protected as trade secrets of Wise Capital, LLC (DTSA, 18 U.S.C. § 1836), with scoring methodology covered by USPTO Provisional Patent Application 64/032,704.

Cascade Dependency Model

Upstream-downstream component relationships propagate risk through the portfolio: upstream failures multiply downstream risk in ways that a flat component-list view cannot capture. Rule set, weights, and propagation logic are trade secret; the concept is auditable from your service-event history.

169

Components Tracked

Across 30 residential appliance categories. Calibrated from ASHRAE 2019 Equipment Life Expectancy, HUD Useful Life Standards, NAHB Component Lifespan Study, BOMA, NEC, and EPA/CDC standards.

U.S. States EDI-Calibrated

Environmental Degradation Index (EDI) factors mapped by zip3 across every U.S. state. The same appliance ages faster in Phoenix than in Anchorage — ForVue accounts for it automatically.

The 4-Step Validation Loop

Predictions don't
evaporate. They get scored.

Every score generated by ForVue is paired with an outcome. When the outcome is recorded, the model learns. This is the difference between deterministic mathematics that improves and black-box AI that drifts.

Prediction Made

ForVue scores a component CRITICAL with a predicted failure window. The prediction is logged with timestamp, scoring inputs, and predicted timeline.

Outcome Logged

When the component is serviced, replaced, or fails, the result is logged in service history. Outcome ties back to the original prediction.

Model Updates

The Bayesian layer incorporates the outcome. Future predictions of similar components improve automatically. The model gets sharper as you use it.

Accuracy Reported

Your accuracy score updates in your dashboard — a living, auditable record. Every number traces back to a logged prediction and outcome.

The Methodology

Weibull-Bayesian.
Patent pending. Deterministic.

The math is generated from your maintenance team's daily use. Adoption is the gatekeeper to score quality.

The scoring engine combines two well-established mathematical frameworks: the Weibull distribution (a survival-analysis curve used in aerospace and reliability engineering since 1951) and Bayesian inference (a probabilistic update mechanism used in any system that learns from outcomes).

📐

Weibull Survival

Each appliance type has a calibrated Weibull failure curve (shape parameter k, scale parameter λ) derived from ASHRAE 2019 Equipment Life Expectancy data and validated against real Wise Capital failures. The curve tells you the probability of survival at any age.

🔄

Bayesian Update

Every logged outcome adjusts the posterior probability for similar components. After enough validation events, the model approaches the true failure distribution of your specific portfolio — not a generic industry average.

📋

Cited Sources

24 CFR Part 968. HUD Notice PIH 2011-7. ASHRAE 2019 Equipment Life Expectancy. NAHB Component Lifespan Study. BOMA. NEC inspection intervals. EPA/CDC indoor air quality. Every input traces to a citable standard.

⚖️

Deterministic Output

Same inputs always produce the same outputs. No randomness, no stochastic sampling, no hidden weights. Any auditor with access to the inputs can recreate any score by hand.

🔐

Patent Pending

USPTO Provisional Patent 64/032,704 (Filed April 8, 2026) — Composite Weibull-Bayesian Predictive Maintenance Scoring System. Sole inventor: Christopher Matthew Wise (Louisville, KY). Foreign filing license granted June 15, 2026. Protected by pending US patent application and DTSA trade secret protection.

📊

Self-Published Accuracy

The dashboard publishes ForVue's diagnostic accuracy live: diagnoses run, resolved via $0 fix, failures pre-flagged, timing error (predicted vs actual year), and per-symptom resolution rate. Computed from your team's logged outcomes — the math sharpens as the team uses the tool. A metric most maintenance platforms don't measure, let alone publish.

🚫

Not Black-Box AI

The risk scores are pure mathematics. AI handles only the language layer — plain-English recommendations and photo parsing. The financial decisions stay with the math.

Every prediction tracked.Every outcome recorded.ForVue tells you how accurate it is.

Predictions don'tevaporate. They get scored.

Weibull-Bayesian.Patent pending. Deterministic.

Math you can audit.Numbers you can defend.

Every prediction tracked.
Every outcome recorded.
ForVue tells you how accurate it is.

Predictions don't
evaporate. They get scored.

Weibull-Bayesian.
Patent pending. Deterministic.

Math you can audit.
Numbers you can defend.