# incrmme
> Compute a moving [mean error][mean-absolute-error] (ME) incrementally.
For a window of size `W`, the [mean error][mean-absolute-error] is defined as
## Usage
```javascript
var incrmme = require( '@stdlib/stats/incr/mme' );
```
#### incrmme( window )
Returns an accumulator `function` which incrementally computes a moving [mean error][mean-absolute-error]. The `window` parameter defines the number of values over which to compute the moving [mean error][mean-absolute-error].
```javascript
var accumulator = incrmme( 3 );
```
#### accumulator( \[x, y] )
If provided input values `x` and `y`, the accumulator function returns an updated [mean error][mean-absolute-error]. If not provided input values `x` and `y`, the accumulator function returns the current [mean error][mean-absolute-error].
```javascript
var accumulator = incrmme( 3 );
var m = accumulator();
// returns null
// Fill the window...
m = accumulator( 2.0, 3.0 ); // [(2.0,3.0)]
// returns 1.0
m = accumulator( -1.0, 4.0 ); // [(2.0,3.0), (-1.0,4.0)]
// returns 3.0
m = accumulator( 3.0, 9.0 ); // [(2.0,3.0), (-1.0,4.0), (3.0,9.0)]
// returns 4.0
// Window begins sliding...
m = accumulator( -7.0, 3.0 ); // [(-1.0,4.0), (3.0,9.0), (-7.0,3.0)]
// returns 7.0
m = accumulator( -5.0, -3.0 ); // [(3.0,9.0), (-7.0,3.0), (-5.0,-3.0)]
// returns 6.0
m = accumulator();
// returns 6.0
```
## Notes
- Input values are **not** type checked. If provided `NaN` or a value which, when used in computations, results in `NaN`, the accumulated value is `NaN` for **at least** `W-1` future invocations. If non-numeric inputs are possible, you are advised to type check and handle accordingly **before** passing the value to the accumulator function.
- As `W` (x,y) pairs are needed to fill the window buffer, the first `W-1` returned values are calculated from smaller sample sizes. Until the window is full, each returned value is calculated from all provided values.
- Be careful when interpreting the [mean error][mean-absolute-error] as errors can cancel. This stated, that errors can cancel makes the [mean error][mean-absolute-error] suitable for measuring the bias in forecasts.
- **Warning**: the [mean error][mean-absolute-error] is scale-dependent and, thus, the measure should **not** be used to make comparisons between datasets having different scales.
## Examples
```javascript
var randu = require( '@stdlib/random/base/randu' );
var incrmme = require( '@stdlib/stats/incr/mme' );
var accumulator;
var v1;
var v2;
var i;
// Initialize an accumulator:
accumulator = incrmme( 5 );
// For each simulated datum, update the moving mean error...
for ( i = 0; i < 100; i++ ) {
v1 = ( randu()*100.0 ) - 50.0;
v2 = ( randu()*100.0 ) - 50.0;
accumulator( v1, v2 );
}
console.log( accumulator() );
```
[mean-absolute-error]: https://en.wikipedia.org/wiki/Mean_absolute_error
