Standard Deviation (Z-Score)
Marigold utilizes statistical variance models to identify systemic, large-scale anomalies (e.g. mass registration clusters) without relying on deterministic rulesets.
The Z-Score Threshold
Unlike duplicate detection (which is a 1:1 mathematical comparison), modules like HIGH_DENSITY rely on evaluating a record against the entire dataset's distribution curve.
Statistical Variance Formula
Z = (x - μ) / σ
Where x is the local concentration (e.g., voters at a single address), μ is the mean concentration across the zip code, and σ is the standard deviation. A Z-score greater than 3.0 indicates a statistically significant anomaly.
Module Integration
Module Integration
When analyzing housing density, Marigold calculates the rolling Z-Score for every residential structure in your encrypted payload. If an address contains 14 voters, but the mean occupancy in that county is 2.1 (with a standard deviation of 0.8), the Z-Score is computed as 14.875.
This immediately triggers a severity: HIGH flag, returning the encrypted Row ID back to your infrastructure for investigation.