awaforge

stats

Apache-2.0Release 1L3

Descriptive statistics — mean, median, quantile, variance, histogram, normalize.

Module stats | Source packages/front/fw/src/io/math/stats.js | Deps none | Worker-safe yes

Basic statistical computations on arrays of numbers. Single-pass where possible (mean, sum); sorting required for median, quantile, iqr. All functions throw on empty arrays or arrays with NaN.

Resolve

const stats = runtime.resolve('stats');
// Returns: { mean, median, mode, geomean, harmean, variance, stddev, range, iqr, mad, min, max, sum, product, quantile, percentile, histogram, zscore, normalize, standardize }

API

Method Signature Returns
mean (arr: number[]) => number Arithmetic mean
median (arr) => number Median value
mode (arr) => number Most frequent value
geomean (arr) => number Geometric mean
harmean (arr) => number Harmonic mean
variance (arr, sample?) => number Population or sample variance
stddev (arr, sample?) => number Standard deviation
range (arr) => number max - min
iqr (arr) => number Q3 - Q1
mad (arr) => number Median absolute deviation
min / max (arr) => number Extreme values
sum / product (arr) => number Sum / product
quantile (arr, q: [0,1]) => number Type 7 quantile (R standard)
percentile (arr, p: [0,100]) => number Alias quantile(arr, p/100)
histogram (arr, bins: number | edges[]) => {edges, counts} Distribution by bins
zscore (value, arr) => number Z-score
normalize (arr, {min?, max?}) => number[] Remap to [min, max]
standardize (arr) => number[] (x - mean) / stddev

Examples

Basic statistics

const stats = runtime.resolve('stats');

const data = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10];

console.log(stats.mean(data));      // 5.5
console.log(stats.median(data));    // 5.5
console.log(stats.variance(data));  // 8.25 (population)
console.log(stats.stddev(data, true)); // ~3.03 (sample)

Quantiles and distribution

const scores = [45, 52, 61, 71, 75, 80, 85, 90, 95, 100];

const q1 = stats.quantile(scores, 0.25);
const q3 = stats.quantile(scores, 0.75);
console.log(`IQR: ${stats.iqr(scores)}`);

const { edges, counts } = stats.histogram(scores, 5);

Normalization

const normalized = stats.normalize([0, 5, 10], { min: 0, max: 1 });
// [0, 0.5, 1]

const standardized = stats.standardize(data);
// mean ≈ 0, stddev ≈ 1

Worker Usage

const worker = fw.createWorker(
    function ({ libs, args }) {
        return {
            mean: libs.stats.mean(args.data),
            stddev: libs.stats.stddev(args.data),
        };
    },
    { dependencies: ['stats'], args: { data: [1, 2, 3, 4, 5] } }
);

Notes

  • quantile Type 7 (R default method, Excel PERCENTILE) — linear interpolation between sorted values.
  • mean([]) and any call on an empty array throws 'stats: empty array' — no silent NaN return.
  • variance(arr, true) divides by n-1 (sample); false (default) divides by n (population).
  • histogram with a bin count: evenly distributed bins over [min, max]. With explicit edges: bins between successive edges.

See also

  • interp — spatial interpolation
  • linalg — linear algebra