EWMA

An exponential weighted moving average for Node.js and the browser!

npm install -g ewma

Usage

Compute the exponential weighted moving average of a series of values. The time at which you insert the value into Ewma is used to compute a weight (recent points are weighted higher). The parameter for defining the convergence speed (like most decay process) is the half-life.

e.g. with a half-life of 10 unit, if you insert 100 at t=0 and 200 at t=10 the ewma will be equal to (200 - 100)/2 = 150 (half of the distance between the new and the old value).

var ewma = new EWMA(halfLifeMs, initialValue, clock)

• halfLifeMs - {Number} parameter representing the speed of convergence
• initialValue - {Number} initial value
• clock - Optional {Number} clock object used to read time, must support

        `Date.now()` style method. Defaults to `Date`.
  

returns an object computing the ewma average

ewma.insert(x)

• x - The next value, ewma will automatically compute the EWMA based on the

    clock difference between this value and the last time `insert` was
    called
  

ewma.reset(x)

• x - Set the EWMA to exactly x.

ewma.value()

Returns the current EWMA value.

Examples

These are generated using a 500ms interval with a half life indicated in the key. For the source code, or to reproduce yourself, check the Example directory.

Contributions

Contributions welcome, please ensure make runs clean.

License

MIT