ML//Time Series//EWMA

Exponentially-weighted moving average: a running average where recent samples count for more and old ones decay geometrically.

Exponentially-weighted moving average: a running average where recent samples count for more and old ones decay geometrically.

Each new value bumps it up; between values it decays. One tunable constant sets the memory length.

A cheap, causal feature for "how active is this stream right now", and an observable stand-in for Hawkes intensity.

The fast clock in a dual-clock feature set (recent density), paired with a slower accumulator.