{"country":"EG","country_name":"Egypt","date_used":"2026-11-17","honest_caveats":["Descriptive, not causal \u2014 high |zscore| != censorship.","Sensitive to data sparsity; some country-days have very few measurements.","period=7 assumes weekly seasonality; flat-pattern countries get small seasonal component."],"interpretation":"residual = observed - (trend + seasonal); zscore = residual / std(residuals) in this country. |zscore| > 2.0 = seasonally anomalous. Higher = bigger break from this country's own weekly rhythm. ORTHOGONAL to /v1/anomaly/dbscan/{cc} \u2014 STL catches own-pattern breaks; DBSCAN catches shape-anomalous days.","is_anomaly":false,"model_meta":{"algorithm":"STL (Cleveland 1990) period=7 robust=True","min_days_for_fit":60,"n_countries_fit":85,"stl_period":7,"trained_at":"2026-10-31T04:45:09.130617Z","version":"v1","window_days":90,"zscore_threshold":2.0},"n_measurements_that_day":5,"observed":0.4,"residual":0.003608,"schema":"voidly-stl-seasonal-anomaly/v1","seasonal":0.041871,"trend":0.354521,"zscore":-0.0245}
