@ai-on-browser/data-analysis-models/lib/model/balanced_histogram.js

Data analysis model package without any dependencies

/**
 * Balanced histogram thresholding
 */
export default class BalancedHistogramThresholding {
	// https://en.wikipedia.org/wiki/Balanced_histogram_thresholding
	/**
	 * @param {number} [minCount] Minimum data count
	 */
	constructor(minCount = 500) {
		this._minCount = minCount
		this._count = 200
	}

	/**
	 * Returns thresholded values.
	 * @param {number[]} x Training data
	 * @returns {(0 | 1)[]} Predicted values
	 */
	predict(x) {
		const max = x.reduce((m, v) => Math.max(m, v), -Infinity)
		const min = x.reduce((m, v) => Math.min(m, v), Infinity)

		const hist = Array(this._count).fill(0)
		const step = (max - min) / this._count
		for (let i = 0; i < x.length; i++) {
			if (x[i] === max) {
				hist[this._count - 1]++
			} else {
				hist[Math.floor((x[i] - min) / step)]++
			}
		}

		let hs = 0,
			he = hist.length - 1
		while (hist[hs] < this._minCount) {
			hs++
			if (hs >= hist.length) {
				throw "'minCount' is too large."
			}
		}
		while (hist[he] < this._minCount) {
			he--
		}
		let hc = Math.ceil((hs + he) / 2)
		let wl = 0
		let wr = 0

		for (let i = hs; i < he + 1; i++) {
			if (i < hc) {
				wl += hist[i]
			} else {
				wr += hist[i]
			}
		}
		while (hs < he) {
			if (wl > wr) {
				wl -= hist[hs++]
			} else {
				wr -= hist[he--]
			}
			const nc = Math.ceil((hs + he) / 2)
			if (nc < hc) {
				wl -= hist[hc - 1]
				wr += hist[hc - 1]
			} else if (nc > hc) {
				wl += hist[hc]
				wr -= hist[hc]
			}
			hc = nc
		}
		this._t = hc * step + min
		return x.map(v => (v < this._t ? 0 : 1))
	}
}