README.md

incrnanmape

Compute the mean absolute percentage error (MAPE) incrementally, ignoring NaN values.

The mean absolute percentage error is defined as

$$\mathop{\mathrm{MAPE}} = \frac{100}{n} \sum_{i=0}^{n-1} \biggl| \frac{a_i - f_i}{a_i} \biggr|$$

where f_i is the forecast value and a_i is the actual value.

Usage

var incrnanmape = require( '@stdlib/stats/incr/nanmape' );

incrnanmape()

Returns an accumulator function which incrementally computes the mean absolute percentage error, ignoring NaN values.

var accumulator = incrnanmape();

accumulator( [f, a] )

If provided input values f and a, the accumulator function returns an updated mean absolute percentage error. If not provided input values f and a, the accumulator function returns the current mean absolute percentage error.

var accumulator = incrnanmape();

var m = accumulator( 2.0, 3.0 );
// returns ~33.33

m = accumulator( 1.0, 4.0 );
// returns ~54.17

m = accumulator( NaN, 4.0 );
// returns ~54.17

m = accumulator( 3.0, 5.0 );
// returns ~49.44

m = accumulator( 3.0, NaN );
// returns ~49.44

m = accumulator( NaN, NaN );
// returns ~49.44

m = accumulator();
// returns ~49.44

Notes

• Input values are not type checked. If non-numeric inputs are possible, you are advised to type check and handle accordingly before passing the value to the accumulator function.

• Warning: the mean absolute percentage error has several shortcomings:

  • The measure is not suitable for intermittent demand patterns (i.e., when a_i is 0).
  • The mean absolute percentage error is not symmetrical, as the measure cannot exceed 100% for forecasts which are too "low" and has no limit for forecasts which are too "high".
  • When used to compare the accuracy of forecast models (e.g., predicting demand), the measure is biased toward forecasts which are too low.

Examples

var randu = require( '@stdlib/random/base/randu' );
var incrnanmape = require( '@stdlib/stats/incr/nanmape' );

var accumulator;
var v1;
var v2;
var i;

// Initialize an accumulator:
accumulator = incrnanmape();

// For each simulated datum, update the mean absolute percentage error...
for ( i = 0; i < 100; i++ ) {
    v1 = ( randu() < 0.2 ) ? NaN : randu() * 100.0;
    v2 = ( randu() < 0.2 ) ? NaN : randu() * 100.0;
    accumulator( v1, v2 );
}
console.log( '\nFinal MAPE: %d\n', accumulator() );