EstimateModel

enum class MayaFlux::Kinesis::Stochastic::EstimateModel : uint8_t

strong

Strategies for characterizing an evolving stream's statistical behavior.

Stochastic generates a sequence with chosen statistical character. Estimate consumes an arriving sequence and characterizes its statistical character as it evolves, since a live stream from a physical source (a sensor, a device report, any per-frame sample) does not carry a fixed, known noise floor, spread, or trend the way a synthetic signal does. Both are stateful processes that evolve over successive calls; this enum plays the role Algorithm plays for Stochastic, selecting how that evolution is computed rather than what it produces.

Enumerator
ROLLING_VARIANCE
EWM_VARIANCE
MEDIAN_ABSOLUTE_DEVIATION
QUIET_PERIOD_FLOOR
TREND

Definition at line 20 of file Estimate.hpp.

                         : uint8_t {
    ROLLING_VARIANCE, // Variance and floor tracked over a fixed recent window
    EWM_VARIANCE, // Variance and floor tracked via exponential weighting, unbounded memory depth
    MEDIAN_ABSOLUTE_DEVIATION, // Floor tracked via MAD, robust to single outlier spikes
    QUIET_PERIOD_FLOOR, // Floor only updates during self-detected low-activity stretches
    TREND // Running linear trend and the variance not explained by it
};