Function detectAnomalies

detectAnomalies(
    series: Float32Array<ArrayBufferLike> | number[],
    o?: CommonOptions & {
        model?: ModelRef;
        range?: [number, number];
        tolerance?: number;
        warmup?: number;
        window?: number;
    },
): Promise<Anomaly[]>
Flag values outside the range the model expected, using rolling one-step forecasts. Batches the rolling windows so long series stay fast.
Parameters
- series: Float32Array<ArrayBufferLike> | number[]
- o: CommonOptions & {
      model?: ModelRef;
      range?: [number, number];
      tolerance?: number;
      warmup?: number;
      window?: number;
  } = {}
  - Optionalmodel?: ModelRef
  - Optionalrange?: [number, number]
    
    Quantiles bounding the expected range. Default [0.1, 0.9].
  - Optionaltolerance?: number
    
    Extra margin on each side, as a fraction of the range's width. Default 0.5; 0 flags about 20% of noisy points.
  - Optionalwarmup?: number
    
    Values used before the first check. Default 64.
  - Optionalwindow?: number
    
    History used for each forecast. Default 512.
Returns Promise<Anomaly[]>
- Defined in helpers/index.ts:325