@visactor/vutils 1.0.25 → 1.0.26

package/cjs/common/regression-lowess.js

@@ -4,153 +4,127 @@ Object.defineProperty(exports, "__esModule", {
     value: !0
 }), exports.regressionLowess = void 0;
 
-const regression_linear_1 = require("./regression-linear"), regression_utils_1 = require("./regression-utils");
-
-function tricube(u) {
-    const uu = Math.abs(u);
-    if (uu >= 1) return 0;
-    const t = 1 - uu * uu * uu;
-    return t * t * t;
-}
+const regression_solver_1 = require("./regression-solver"), median_1 = require("./median"), regression_utils_1 = require("./regression-utils");
 
 /**
- * Stratified sampling to reduce data size while preserving distribution
- */ function stratifiedSample(sortedData, maxSamples) {
-    const n = sortedData.length;
-    if (n <= maxSamples) return sortedData;
-    const sampled = [], step = n / maxSamples;
-    // More aggressive sampling - use exact step size
-        for (let i = 0; i < maxSamples; i++) {
-        const idx = Math.min(Math.floor(i * step), n - 1);
-        sampled.push(sortedData[idx]);
+ * 局部线性 LOWESS。默认全量观测、span=2/3、三次稳健重加权。
+ * delta 显式插值加速；maxSamples 仅保留显式选择的历史抽样用途。
+ * predict 与 evaluateGrid 查询构建时固定的同一模型，不触发再拟合。
+ */
+function regressionLowess(data, x = (d => d.x), y = (d => d.y), options = {}) {
+    var _a, _b, _c, _d, _e;
+    const span = null !== (_a = options.span) && void 0 !== _a ? _a : 2 / 3, degree = null !== (_b = options.degree) && void 0 !== _b ? _b : 1, iterations = null !== (_c = options.iterations) && void 0 !== _c ? _c : 3, delta = null !== (_d = options.delta) && void 0 !== _d ? _d : 0, alpha = null !== (_e = options.alpha) && void 0 !== _e ? _e : .05;
+    if (!(span > 0 && span <= 1) || !Number.isInteger(iterations) || iterations < 0 || !Number.isFinite(delta) || delta < 0 || 0 !== degree && 1 !== degree || null != options.maxSamples && (!Number.isInteger(options.maxSamples) || options.maxSamples < 2)) throw new RangeError("Invalid LOWESS span, degree, iterations, delta or maxSamples");
+    let pts = (0, regression_solver_1.regressionPoints)(data, x, y).sort(((a, b) => a.x - b.x || a.y - b.y));
+    if (null != options.maxSamples && pts.length > options.maxSamples) {
+        const full = pts, count = options.maxSamples;
+        pts = Array.from({
+            length: count
+        }, ((_, i) => full[Math.floor(i * (full.length - 1) / (count - 1))]));
     }
-    return sampled;
-}
-
-/**
- * Simple lowess implementation (univariate x)
- * options:
- * - span: fraction of points used in local regression (0,1]
- * - degree: 0 (constant) or 1 (linear)
- * - iterations: number of robustifying iterations
- * - alpha: confidence level for CI
- * - maxSamples: maximum number of points to use (default: 1000 for fast processing)
- */ function regressionLowess(data, x = (d => d.x), y = (d => d.y), options = {}) {
-    var _a;
-    const span = options.span || .3, degree = 0 === options.degree ? 0 : 1, alpha = null !== (_a = options.alpha) && void 0 !== _a ? _a : .05, iterations = null == options.iterations ? 2 : options.iterations, maxSamples = options.maxSamples || 1e3, rawPoints = [];
-    (0, regression_linear_1.visitPoints)(data, x, y, ((dx, dy, index) => {
-        rawPoints.push({
-            x: dx,
-            y: dy,
-            index: index
-        });
-    })), rawPoints.sort(((a, b) => a.x - b.x));
-    // Apply sampling if needed
-    const sampledPoints = stratifiedSample(rawPoints, maxSamples), n = sampledPoints.length, ptsX = new Array(n), ptsY = new Array(n);
-    for (let i = 0; i < n; i++) ptsX[i] = sampledPoints[i].x, ptsY[i] = sampledPoints[i].y;
-    /**
-     * Optimized predictSingle using binary search on pre-sorted data
-     */    function predictSingle(x0, robustWeights) {
-        if (0 === n) return 0;
-        // Binary search to find insertion point
-                let left = 0, right = n;
+    const n = pts.length, ptsX = pts.map((p => p.x)), ptsY = pts.map((p => p.y)), basis = (0, 
+    regression_solver_1.regressionBasis)(pts), normalizedX = ptsX.map((v => (0, regression_solver_1.normalizeRegression)(v, basis.xCenter, basis.xScale))), normalizedY = ptsY.map((v => (0, 
+    regression_solver_1.normalizeRegression)(v, basis.yCenter, basis.yScale))), count = Math.min(n, Math.max(2, Math.floor(span * n + 1e-10)));
+    let robustWeights;
+    function exact(x0) {
+        if (!n) return 0;
+        const query = (0, regression_solver_1.normalizeRegression)(x0, basis.xCenter, basis.xScale);
+        // 找到连续的最近邻窗口；相等距离选择较小的 x（ties 的 y 已排序）。
+                let left = 0, right = n - count;
         for (;left < right; ) {
-            const mid = left + right >> 1;
-            ptsX[mid] < x0 ? left = mid + 1 : right = mid;
+            const mid = left + right >>> 1, lowerDistance = x0 - ptsX[mid], upperDistance = ptsX[mid + count] - x0;
+            (Number.isFinite(lowerDistance) && Number.isFinite(upperDistance) ? lowerDistance > upperDistance : query - normalizedX[mid] > normalizedX[mid + count] - query) ? left = mid + 1 : right = mid;
         }
-        const m = Math.min(n, Math.max(2, Math.floor(span * n)));
-        // Expand by distance: equal numbers on either side need not be the nearest points.
-                let lo = left - 1, hi = left;
-        for (let count = 0; count < m; count++) lo >= 0 && (hi >= n || x0 - ptsX[lo] <= ptsX[hi] - x0) ? lo-- : hi++;
-        const actualStart = lo + 1, end = hi;
-        // Find max distance and compute weights in single pass
-        let maxDist = 0;
-        const windowSize = end - actualStart, distances = new Array(windowSize);
-        for (let i = actualStart; i < end; i++) {
-            const dist = Math.abs(ptsX[i] - x0);
-            distances[i - actualStart] = dist, dist > maxDist && (maxDist = dist);
+        let start = left, end = left + count;
+        if (ptsX[start] === ptsX[end - 1]) {
+            for (;start > 0 && ptsX[start - 1] === ptsX[start]; ) start--;
+            for (;end < n && ptsX[end] === ptsX[start]; ) end++;
         }
-        // Compute weights using pre-calculated distances
-                let sumw = 0;
-        const w = new Array(windowSize);
-        for (let i = 0; i < windowSize; i++) {
-            let wi = tricube(0 === maxDist ? 0 : distances[i] / maxDist);
-            robustWeights && null != robustWeights[actualStart + i] && (wi *= robustWeights[actualStart + i]), 
-            w[i] = wi, sumw += wi;
+        const rawRadius = Math.max(Math.abs(ptsX[start] - x0), Math.abs(ptsX[end - 1] - x0)), scaled = !Number.isFinite(rawRadius), radius = scaled ? Math.max(Math.abs(normalizedX[start] - query), Math.abs(normalizedX[end - 1] - query)) : rawRadius;
+        // 局部响应坐标不能由远处、未参与该邻域的观测决定精度。
+        let yMin = 1 / 0, yMax = -1 / 0;
+        for (let i = start; i < end; i++) yMin = Math.min(yMin, ptsY[i]), yMax = Math.max(yMax, ptsY[i]);
+        const yCenter = yMin / 2 + yMax / 2, yScale = Math.max(Math.abs(yMin - yCenter), Math.abs(yMax - yCenter)) || 1;
+        let sw = 0, sx = 0, sy = 0, sxx = 0, sxy = 0;
+        for (let i = start; i < end; i++) {
+            const z = 0 === radius ? 0 : (scaled ? normalizedX[i] - query : ptsX[i] - x0) / radius, u = Math.abs(z), t = u >= 1 ? 0 : 1 - u * u * u, w = t * t * t * (robustWeights ? robustWeights[i] : 1);
+            sw += w, sx += w * z;
+            const response = (0, regression_solver_1.normalizeRegression)(ptsY[i], yCenter, yScale);
+            sy += w * response, sxx += w * z * z, sxy += w * z * response;
         }
-        if (0 === sumw) {
-            // Fall back to the nearest y, preserving input order for equal distances.
-            let nearestIdx = 0, nearestDistance = Math.abs(ptsX[0] - x0);
-            for (let i = 1; i < n; i++) {
-                const distance = Math.abs(ptsX[i] - x0);
-                (distance < nearestDistance || distance === nearestDistance && sampledPoints[i].index < sampledPoints[nearestIdx].index) && (nearestIdx = i, 
-                nearestDistance = distance);
+        if (0 === sw) {
+            // 核边界零权重时使用最近有效观测的加权常数；等距离取均值。
+            let distance = 1 / 0, weight = 0, value = 0;
+            for (let i = 0; i < n; i++) {
+                const w = robustWeights ? robustWeights[i] : 1;
+                if (0 === w) continue;
+                const d = Math.abs(scaled ? normalizedX[i] - query : ptsX[i] - x0);
+                d < distance ? (distance = d, weight = w, value = ptsY[i]) : d === distance && (value = value * (weight / (weight + w)) + ptsY[i] * (w / (weight + w)), 
+                weight += w);
             }
-            return ptsY[nearestIdx];
+            if (0 === weight) throw new RangeError("LOWESS has no effective local observations");
+            return value;
         }
-        if (0 === degree) {
-            let s = 0;
-            for (let i = 0; i < w.length; i++) s += w[i] * ptsY[actualStart + i];
-            return s / sumw;
+        const mx = sx / sw, my = sy / sw, variance = sxx / sw - mx * mx;
+        return yCenter + yScale * (my - (0 === degree || variance <= 32 * Number.EPSILON ? 0 : (sxy / sw - mx * my) / variance) * mx);
+    }
+    // delta 不删除观测。只减少局部回归求值位置，全部残差参与重加权。
+        const anchors = [];
+    for (let i = 0; i < n && (anchors.push(i), i !== n - 1); ) {
+        let next = i + 1;
+        if (delta > 0) for (;next + 1 < n && ptsX[next + 1] - ptsX[i] <= delta; ) next++;
+        // 重复 x 一起使用同一预测；端点不能丢失。
+                for (;next + 1 < n && ptsX[next + 1] === ptsX[next]; ) next++;
+        i = next;
+    }
+    let fits = new Array(n);
+    function fitObservations() {
+        const out = new Array(n);
+        for (let a = 0; a < anchors.length; a++) {
+            const index = anchors[a];
+            if (out[index] = exact(ptsX[index]), 0 === a) continue;
+            const previous = anchors[a - 1], rawWidth = ptsX[index] - ptsX[previous], width = Number.isFinite(rawWidth) ? rawWidth : normalizedX[index] - normalizedX[previous];
+            for (let i = previous + 1; i < index; i++) {
+                const t = 0 === width ? 0 : (Number.isFinite(rawWidth) ? ptsX[i] - ptsX[previous] : normalizedX[i] - normalizedX[previous]) / width;
+                out[i] = out[previous] * (1 - t) + out[index] * t;
+            }
         }
-        // weighted linear regression on local points
-                let sw = 0, sx = 0, sy = 0, sxx = 0, sxy = 0;
-        for (let i = actualStart; i < end; i++) {
-            const idx = i - actualStart, xi = ptsX[i], yi = ptsY[i], wi = w[idx];
-            sw += wi, sx += wi * xi, sy += wi * yi, sxx += wi * xi * xi, sxy += wi * xi * yi;
+        return out;
+    }
+    for (let iteration = 0; iteration <= iterations && (fits = fitObservations(), iteration !== iterations); iteration++) {
+        const residuals = fits.map(((v, i) => {
+            const residual = ptsY[i] - v;
+            return Number.isFinite(residual) ? Math.abs(residual / basis.yScale) : Math.abs(normalizedY[i] - (0, 
+            regression_solver_1.normalizeRegression)(v, basis.yCenter, basis.yScale));
+        })), med = (0, median_1.median)(residuals.slice()), currentFits = fits;
+        robustWeights = residuals.map(((r, i) => {
+            if (0 === med) {
+                return r <= 64 * Number.EPSILON * (Math.max(Math.abs(ptsY[i]), Math.abs(currentFits[i])) / basis.yScale) ? 1 : 0;
+            }
+            const u = r / (6 * med);
+            return u >= 1 ? 0 : (1 - u * u) ** 2;
+        }));
+    }
+    function predictSingle(x0) {
+        if (0 === delta || n < 2 || x0 < ptsX[0] || x0 > ptsX[n - 1]) return exact(x0);
+        let lo = 0, hi = n - 1;
+        for (;hi - lo > 1; ) {
+            const mid = lo + hi >>> 1;
+            ptsX[mid] <= x0 ? lo = mid : hi = mid;
         }
-        const meanX = sx / sw, meanY = sy / sw, denom = sxx - sx * meanX, slope = Math.abs(denom) < 1e-12 ? 0 : (sxy - sx * meanY) / denom;
-        return meanY - slope * meanX + slope * x0;
+        if (ptsX[hi] === ptsX[lo]) return fits[lo];
+        const width = ptsX[hi] - ptsX[lo], t = Number.isFinite(width) ? (x0 - ptsX[lo]) / width : ((0, 
+        regression_solver_1.normalizeRegression)(x0, basis.xCenter, basis.xScale) - normalizedX[lo]) / (normalizedX[hi] - normalizedX[lo]);
+        return fits[lo] * (1 - t) + fits[hi] * t;
     }
     function predict(x0) {
-        if (Array.isArray(x0)) {
-            const len = x0.length, out = new Array(len);
-            for (let i = 0; i < len; i++) out[i] = predictSingle(x0[i]);
-            return out;
-        }
-        return predictSingle(x0);
+        return Array.isArray(x0) ? x0.map(predictSingle) : predictSingle(x0);
     }
     return {
         predict: predict,
         evaluate: predict,
         evaluateGrid: function(N) {
-            if (N <= 0) return [];
-            if (0 === n) return [];
-            const out = new Array(N), min = ptsX[0], max = ptsX[n - 1];
-            if (min === max) {
-                const v = predictSingle(min);
-                for (let i = 0; i < N; i++) out[i] = {
-                    x: min,
-                    y: v
-                };
-                return out;
-            }
-            const step = (max - min) / (N - 1);
-            // optionally add robust iterations
-                        let robustWeights;
-            // Disable robustness iterations for large datasets to improve performance
-            // Users can override by setting iterations explicitly
-                        const effectiveIterations = null != options.iterations ? iterations : n > 500 ? 0 : iterations;
-            if (effectiveIterations > 0) for (let iter = 0; iter < effectiveIterations; iter++) {
-                // compute fits - pre-allocate arrays
-                const fits = new Array(n), res = new Array(n);
-                for (let i = 0; i < n; i++) fits[i] = predictSingle(ptsX[i], robustWeights), res[i] = Math.abs(ptsY[i] - fits[i]);
-                // median absolute deviation
-                                const med = res.slice().sort(((a, b) => a - b))[Math.floor(n / 2)] || 0;
-                robustWeights = new Array(n);
-                for (let i = 0; i < n; i++) {
-                    const u = 0 === med ? 0 : res[i] / (6 * med), w = Math.abs(u) >= 1 ? 0 : (1 - u * u) * (1 - u * u);
-                    robustWeights[i] = w;
-                }
-            }
-            for (let i = 0; i < N; i++) {
-                const px = i === N - 1 ? max : min + step * i;
-                out[i] = {
-                    x: px,
-                    y: predictSingle(px, robustWeights)
-                };
-            }
-            return out;
+            return (0, regression_solver_1.regressionGrid)(ptsX[0], ptsX[n - 1], N, predictSingle);
         },
         confidenceInterval: function(N = 50) {
             const out = [];
@@ -190,7 +164,12 @@ function tricube(u) {
                 });
             }
             return out;
-        }
+        },
+        // 节点值按 x 排序，重复观测仍保持其统计权重。
+        fitted: pts.map(((p, i) => ({
+            x: p.x,
+            y: fits[i]
+        })))
     };
 }
 

package/cjs/common/regression-lowess.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../src/common/regression-lowess.ts"],"names":[],"mappings":";;;AAAA,2DAAkD;AAClD,yDAK4B;AAQ5B,SAAS,OAAO,CAAC,CAAS;IACxB,MAAM,EAAE,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;IACvB,IAAI,EAAE,IAAI,CAAC,EAAE;QACX,OAAO,CAAC,CAAC;KACV;IACD,MAAM,CAAC,GAAG,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,CAAC;IAC3B,OAAO,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;AACnB,CAAC;AAED;;GAEG;AACH,SAAS,gBAAgB,CAAC,UAAyB,EAAE,UAAkB;IACrE,MAAM,CAAC,GAAG,UAAU,CAAC,MAAM,CAAC;IAC5B,IAAI,CAAC,IAAI,UAAU,EAAE;QACnB,OAAO,UAAU,CAAC;KACnB;IAED,MAAM,OAAO,GAAkB,EAAE,CAAC;IAElC,iDAAiD;IACjD,MAAM,IAAI,GAAG,CAAC,GAAG,UAAU,CAAC;IAE5B,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,UAAU,EAAE,CAAC,EAAE,EAAE;QACnC,MAAM,GAAG,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,GAAG,IAAI,CAAC,EAAE,CAAC,GAAG,CAAC,CAAC,CAAC;QAClD,OAAO,CAAC,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC;KAC/B;IAED,OAAO,OAAO,CAAC;AACjB,CAAC;AAED;;;;;;;;GAQG;AACH,SAAgB,gBAAgB,CAC9B,IAAW,EACX,IAAwB,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,EAChC,IAAwB,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,EAChC,UAAuG,EAAE;;IAEzG,MAAM,IAAI,GAAG,OAAO,CAAC,IAAI,IAAI,GAAG,CAAC;IACjC,MAAM,MAAM,GAAG,OAAO,CAAC,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IAC5C,MAAM,KAAK,GAAG,MAAA,OAAO,CAAC,KAAK,mCAAI,IAAI,CAAC;IACpC,MAAM,UAAU,GAAG,OAAO,CAAC,UAAU,IAAI,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,UAAU,CAAC;IACvE,MAAM,UAAU,GAAG,OAAO,CAAC,UAAU,IAAI,IAAI,CAAC;IAE9C,6BAA6B;IAC7B,MAAM,SAAS,GAAkB,EAAE,CAAC;IACpC,IAAA,+BAAW,EAAC,IAAI,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,EAAE,EAAE,EAAE,KAAK,EAAE,EAAE;QACxC,SAAS,CAAC,IAAI,CAAC,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,KAAK,EAAE,CAAC,CAAC;IAC1C,CAAC,CAAC,CAAC;IAEH,SAAS,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;IAEpC,2BAA2B;IAC3B,MAAM,aAAa,GAAG,gBAAgB,CAAC,SAAS,EAAE,UAAU,CAAC,CAAC;IAE9D,MAAM,CAAC,GAAG,aAAa,CAAC,MAAM,CAAC;IAC/B,MAAM,IAAI,GAAa,IAAI,KAAK,CAAC,CAAC,CAAC,CAAC;IACpC,MAAM,IAAI,GAAa,IAAI,KAAK,CAAC,CAAC,CAAC,CAAC;IAEpC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE;QAC1B,IAAI,CAAC,CAAC,CAAC,GAAG,aAAa,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QAC7B,IAAI,CAAC,CAAC,CAAC,GAAG,aAAa,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;KAC9B;IAED;;OAEG;IACH,SAAS,aAAa,CAAC,EAAU,EAAE,aAAwB;QACzD,IAAI,CAAC,KAAK,CAAC,EAAE;YACX,OAAO,CAAC,CAAC;SACV;QAED,wCAAwC;QACxC,IAAI,IAAI,GAAG,CAAC,CAAC;QACb,IAAI,KAAK,GAAG,CAAC,CAAC;QACd,OAAO,IAAI,GAAG,KAAK,EAAE;YACnB,MAAM,GAAG,GAAG,CAAC,IAAI,GAAG,KAAK,CAAC,IAAI,CAAC,CAAC;YAChC,IAAI,IAAI,CAAC,GAAG,CAAC,GAAG,EAAE,EAAE;gBAClB,IAAI,GAAG,GAAG,GAAG,CAAC,CAAC;aAChB;iBAAM;gBACL,KAAK,GAAG,GAAG,CAAC;aACb;SACF;QAED,MAAM,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,IAAI,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;QAEzD,mFAAmF;QACnF,IAAI,EAAE,GAAG,IAAI,GAAG,CAAC,CAAC;QAClB,IAAI,EAAE,GAAG,IAAI,CAAC;QACd,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,GAAG,CAAC,EAAE,KAAK,EAAE,EAAE;YACtC,IAAI,EAAE,IAAI,CAAC,IAAI,CAAC,EAAE,IAAI,CAAC,IAAI,EAAE,GAAG,IAAI,CAAC,EAAE,CAAC,IAAI,IAAI,CAAC,EAAE,CAAC,GAAG,EAAE,CAAC,EAAE;gBAC1D,EAAE,EAAE,CAAC;aACN;iBAAM;gBACL,EAAE,EAAE,CAAC;aACN;SACF;QACD,MAAM,WAAW,GAAG,EAAE,GAAG,CAAC,CAAC;QAC3B,MAAM,GAAG,GAAG,EAAE,CAAC;QAEf,uDAAuD;QACvD,IAAI,OAAO,GAAG,CAAC,CAAC;QAChB,MAAM,UAAU,GAAG,GAAG,GAAG,WAAW,CAAC;QACrC,MAAM,SAAS,GAAa,IAAI,KAAK,CAAC,UAAU,CAAC,CAAC;QAElD,KAAK,IAAI,CAAC,GAAG,WAAW,EAAE,CAAC,GAAG,GAAG,EAAE,CAAC,EAAE,EAAE;YACtC,MAAM,IAAI,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC,CAAC;YACpC,SAAS,CAAC,CAAC,GAAG,WAAW,CAAC,GAAG,IAAI,CAAC;YAClC,IAAI,IAAI,GAAG,OAAO,EAAE;gBAClB,OAAO,GAAG,IAAI,CAAC;aAChB;SACF;QAED,iDAAiD;QACjD,IAAI,IAAI,GAAG,CAAC,CAAC;QACb,MAAM,CAAC,GAAa,IAAI,KAAK,CAAC,UAAU,CAAC,CAAC;QAC1C,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,UAAU,EAAE,CAAC,EAAE,EAAE;YACnC,MAAM,CAAC,GAAG,OAAO,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,GAAG,OAAO,CAAC;YACrD,IAAI,EAAE,GAAG,OAAO,CAAC,CAAC,CAAC,CAAC;YACpB,IAAI,aAAa,IAAI,aAAa,CAAC,WAAW,GAAG,CAAC,CAAC,IAAI,IAAI,EAAE;gBAC3D,EAAE,IAAI,aAAa,CAAC,WAAW,GAAG,CAAC,CAAC,CAAC;aACtC;YACD,CAAC,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC;YACV,IAAI,IAAI,EAAE,CAAC;SACZ;QAED,IAAI,IAAI,KAAK,CAAC,EAAE;YACd,0EAA0E;YAC1E,IAAI,UAAU,GAAG,CAAC,CAAC;YACnB,IAAI,eAAe,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC,CAAC;YAC7C,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE;gBAC1B,MAAM,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC,CAAC;gBACxC,IACE,QAAQ,GAAG,eAAe;oBAC1B,CAAC,QAAQ,KAAK,eAAe,IAAI,aAAa,CAAC,CAAC,CAAC,CAAC,KAAK,GAAG,aAAa,CAAC,UAAU,CAAC,CAAC,KAAK,CAAC,EAC1F;oBACA,UAAU,GAAG,CAAC,CAAC;oBACf,eAAe,GAAG,QAAQ,CAAC;iBAC5B;aACF;YACD,OAAO,IAAI,CAAC,UAAU,CAAC,CAAC;SACzB;QAED,IAAI,MAAM,KAAK,CAAC,EAAE;YAChB,IAAI,CAAC,GAAG,CAAC,CAAC;YACV,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;gBACjC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,GAAG,IAAI,CAAC,WAAW,GAAG,CAAC,CAAC,CAAC;aACnC;YACD,OAAO,CAAC,GAAG,IAAI,CAAC;SACjB;QAED,6CAA6C;QAC7C,IAAI,EAAE,GAAG,CAAC,CAAC;QACX,IAAI,EAAE,GAAG,CAAC,CAAC;QACX,IAAI,EAAE,GAAG,CAAC,CAAC;QACX,IAAI,GAAG,GAAG,CAAC,CAAC;QACZ,IAAI,GAAG,GAAG,CAAC,CAAC;QACZ,KAAK,IAAI,CAAC,GAAG,WAAW,EAAE,CAAC,GAAG,GAAG,EAAE,CAAC,EAAE,EAAE;YACtC,MAAM,GAAG,GAAG,CAAC,GAAG,WAAW,CAAC;YAC5B,MAAM,EAAE,GAAG,IAAI,CAAC,CAAC,CAAC,CAAC;YACnB,MAAM,EAAE,GAAG,IAAI,CAAC,CAAC,CAAC,CAAC;YACnB,MAAM,EAAE,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC;YAClB,EAAE,IAAI,EAAE,CAAC;YACT,EAAE,IAAI,EAAE,GAAG,EAAE,CAAC;YACd,EAAE,IAAI,EAAE,GAAG,EAAE,CAAC;YACd,GAAG,IAAI,EAAE,GAAG,EAAE,GAAG,EAAE,CAAC;YACpB,GAAG,IAAI,EAAE,GAAG,EAAE,GAAG,EAAE,CAAC;SACrB;QAED,MAAM,KAAK,GAAG,EAAE,GAAG,EAAE,CAAC;QACtB,MAAM,KAAK,GAAG,EAAE,GAAG,EAAE,CAAC;QACtB,MAAM,KAAK,GAAG,GAAG,GAAG,EAAE,GAAG,KAAK,CAAC;QAC/B,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,GAAG,EAAE,GAAG,KAAK,CAAC,GAAG,KAAK,CAAC;QACvE,MAAM,SAAS,GAAG,KAAK,GAAG,KAAK,GAAG,KAAK,CAAC;QACxC,OAAO,SAAS,GAAG,KAAK,GAAG,EAAE,CAAC;IAChC,CAAC;IAED,SAAS,OAAO,CAAC,EAAqB;QACpC,IAAI,KAAK,CAAC,OAAO,CAAC,EAAE,CAAC,EAAE;YACrB,MAAM,GAAG,GAAG,EAAE,CAAC,MAAM,CAAC;YACtB,MAAM,GAAG,GAAa,IAAI,KAAK,CAAC,GAAG,CAAC,CAAC;YACrC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,GAAG,EAAE,CAAC,EAAE,EAAE;gBAC5B,GAAG,CAAC,CAAC,CAAC,GAAG,aAAa,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC;aAC/B;YACD,OAAO,GAAG,CAAC;SACZ;QACD,OAAO,aAAa,CAAC,EAAY,CAAC,CAAC;IACrC,CAAC;IAED,SAAS,YAAY,CAAC,CAAS;QAC7B,IAAI,CAAC,IAAI,CAAC,EAAE;YACV,OAAO,EAAE,CAAC;SACX;QACD,IAAI,CAAC,KAAK,CAAC,EAAE;YACX,OAAO,EAAE,CAAC;SACX;QAED,MAAM,GAAG,GAA+B,IAAI,KAAK,CAAC,CAAC,CAAC,CAAC;QACrD,MAAM,GAAG,GAAG,IAAI,CAAC,CAAC,CAAC,CAAC;QACpB,MAAM,GAAG,GAAG,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;QAExB,IAAI,GAAG,KAAK,GAAG,EAAE;YACf,MAAM,CAAC,GAAG,aAAa,CAAC,GAAG,CAAC,CAAC;YAC7B,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE;gBAC1B,GAAG,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC;aAC3B;YACD,OAAO,GAAG,CAAC;SACZ;QACD,MAAM,IAAI,GAAG,CAAC,GAAG,GAAG,GAAG,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;QAEnC,mCAAmC;QACnC,IAAI,aAAmC,CAAC;QAExC,0EAA0E;QAC1E,sDAAsD;QACtD,MAAM,mBAAmB,GAAG,OAAO,CAAC,UAAU,IAAI,IAAI,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,GAAG,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,UAAU,CAAC;QAE/F,IAAI,mBAAmB,GAAG,CAAC,EAAE;YAC3B,KAAK,IAAI,IAAI,GAAG,CAAC,EAAE,IAAI,GAAG,mBAAmB,EAAE,IAAI,EAAE,EAAE;gBACrD,qCAAqC;gBACrC,MAAM,IAAI,GAAa,IAAI,KAAK,CAAC,CAAC,CAAC,CAAC;gBACpC,MAAM,GAAG,GAAa,IAAI,KAAK,CAAC,CAAC,CAAC,CAAC;gBAEnC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE;oBAC1B,IAAI,CAAC,CAAC,CAAC,GAAG,aAAa,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,aAAa,CAAC,CAAC;oBAChD,GAAG,CAAC,CAAC,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;iBACtC;gBAED,4BAA4B;gBAC5B,MAAM,SAAS,GAAG,GAAG,CAAC,KAAK,EAAE,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;gBACpD,MAAM,GAAG,GAAG,SAAS,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;gBAC9C,aAAa,GAAG,IAAI,KAAK,CAAC,CAAC,CAAC,CAAC;gBAC7B,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE;oBAC1B,MAAM,CAAC,GAAG,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,GAAG,CAAC,CAAC;oBAC7C,MAAM,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC;oBAC3D,aAAa,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC;iBACtB;aACF;SACF;QAED,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE;YAC1B,MAAM,EAAE,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,GAAG,IAAI,GAAG,CAAC,CAAC;YAC9C,GAAG,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,aAAa,CAAC,EAAE,EAAE,aAAa,CAAC,EAAE,CAAC;SACzD;QACD,OAAO,GAAG,CAAC;IACb,CAAC;IAED,SAAS,kBAAkB,CAAC,IAAY,EAAE;QACxC,MAAM,GAAG,GAAsG,EAAE,CAAC;QAElH,IAAI,CAAC,IAAI,CAAC,EAAE;YACV,OAAO,GAAG,CAAC;SACZ;QACD,IAAI,CAAC,KAAK,CAAC,EAAE;YACX,OAAO,GAAG,CAAC;SACZ;QAED,oCAAoC;QACpC,MAAM,GAAG,GAAG,IAAI,CAAC,CAAC,CAAC,CAAC;QACpB,MAAM,GAAG,GAAG,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;QAExB,IAAI,GAAG,KAAK,SAAS,IAAI,GAAG,KAAK,SAAS,IAAI,GAAG,KAAK,QAAQ,IAAI,GAAG,KAAK,CAAC,QAAQ,EAAE;YACnF,OAAO,GAAG,CAAC;SACZ;QAED,MAAM,KAAK,GAAG,IAAA,4CAAyB,EAAC,IAAI,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAU,EAAE,EAAE,CAAC,aAAa,CAAC,EAAE,CAAC,CAAC,CAAC;QACvF,IAAI,KAAK,CAAC,CAAC,KAAK,CAAC,EAAE;YACjB,OAAO,GAAG,CAAC;SACZ;QAED,MAAM,CAAC,GAAG,IAAA,0CAAuB,EAAC,KAAK,CAAC,CAAC;QACzC,IAAI,KAAK,CAAC,GAAG,KAAK,KAAK,CAAC,GAAG,EAAE;YAC3B,MAAM,CAAC,GAAG,aAAa,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;YACnC,MAAM,IAAI,GAAG,IAAA,8BAAW,EAAC,KAAK,CAAC,GAAG,EAAE,KAAK,CAAC,CAAC;YAC3C,MAAM,aAAa,GAAG,IAAA,sCAAmB,EAAC,CAAC,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;YAC1D,MAAM,aAAa,GAAG,IAAA,sCAAmB,EAAC,CAAC,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;YAC1D,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE;gBAC1B,GAAG,CAAC,IAAI,CAAC;oBACP,CAAC,EAAE,KAAK,CAAC,GAAG;oBACZ,IAAI,EAAE,CAAC;oBACP,KAAK,EAAE,CAAC,GAAG,aAAa;oBACxB,KAAK,EAAE,CAAC,GAAG,aAAa;oBACxB,SAAS,EAAE,CAAC,GAAG,aAAa;oBAC5B,SAAS,EAAE,CAAC,GAAG,aAAa;iBAC7B,CAAC,CAAC;aACJ;YACD,OAAO,GAAG,CAAC;SACZ;QAED,MAAM,IAAI,GAAG,CAAC,GAAG,GAAG,GAAG,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;QACnC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE;YAC1B,MAAM,EAAE,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,GAAG,IAAI,GAAG,CAAC,CAAC;YAC9C,MAAM,EAAE,GAAG,aAAa,CAAC,EAAE,CAAC,CAAC;YAC7B,MAAM,IAAI,GAAG,IAAA,8BAAW,EAAC,EAAE,EAAE,KAAK,CAAC,CAAC;YACpC,MAAM,aAAa,GAAG,IAAA,sCAAmB,EAAC,CAAC,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;YAC1D,MAAM,aAAa,GAAG,IAAA,sCAAmB,EAAC,CAAC,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;YAC1D,GAAG,CAAC,IAAI,CAAC;gBACP,CAAC,EAAE,EAAE;gBACL,IAAI,EAAE,EAAE;gBACR,KAAK,EAAE,EAAE,GAAG,aAAa;gBACzB,KAAK,EAAE,EAAE,GAAG,aAAa;gBACzB,SAAS,EAAE,EAAE,GAAG,aAAa;gBAC7B,SAAS,EAAE,EAAE,GAAG,aAAa;aAC9B,CAAC,CAAC;SACJ;QACD,OAAO,GAAG,CAAC;IACb,CAAC;IAED,OAAO;QACL,OAAO;QACP,QAAQ,EAAE,OAAc;QACxB,YAAY;QACZ,kBAAkB;KACnB,CAAC;AACJ,CAAC;AAzRD,4CAyRC;AAED,kBAAe,gBAAgB,CAAC","file":"regression-lowess.js","sourcesContent":["import { visitPoints } from './regression-linear';\nimport {\n  computeLinearCIComponents,\n  confidenceCriticalValue,\n  confidenceHalfWidth,\n  stdErrorsAt\n} from './regression-utils';\n\ninterface LowessPoint {\n  x: number;\n  y: number;\n  index: number;\n}\n\nfunction tricube(u: number) {\n  const uu = Math.abs(u);\n  if (uu >= 1) {\n    return 0;\n  }\n  const t = 1 - uu * uu * uu;\n  return t * t * t;\n}\n\n/**\n * Stratified sampling to reduce data size while preserving distribution\n */\nfunction stratifiedSample(sortedData: LowessPoint[], maxSamples: number) {\n  const n = sortedData.length;\n  if (n <= maxSamples) {\n    return sortedData;\n  }\n\n  const sampled: LowessPoint[] = [];\n\n  // More aggressive sampling - use exact step size\n  const step = n / maxSamples;\n\n  for (let i = 0; i < maxSamples; i++) {\n    const idx = Math.min(Math.floor(i * step), n - 1);\n    sampled.push(sortedData[idx]);\n  }\n\n  return sampled;\n}\n\n/**\n * Simple lowess implementation (univariate x)\n * options:\n * - span: fraction of points used in local regression (0,1]\n * - degree: 0 (constant) or 1 (linear)\n * - iterations: number of robustifying iterations\n * - alpha: confidence level for CI\n * - maxSamples: maximum number of points to use (default: 1000 for fast processing)\n */\nexport function regressionLowess(\n  data: any[],\n  x: (d: any) => number = d => d.x,\n  y: (d: any) => number = d => d.y,\n  options: { span?: number; degree?: 1 | 0; iterations?: number; alpha?: number; maxSamples?: number } = {}\n) {\n  const span = options.span || 0.3;\n  const degree = options.degree === 0 ? 0 : 1;\n  const alpha = options.alpha ?? 0.05;\n  const iterations = options.iterations == null ? 2 : options.iterations;\n  const maxSamples = options.maxSamples || 1000;\n\n  // Collect and sort data by x\n  const rawPoints: LowessPoint[] = [];\n  visitPoints(data, x, y, (dx, dy, index) => {\n    rawPoints.push({ x: dx, y: dy, index });\n  });\n\n  rawPoints.sort((a, b) => a.x - b.x);\n\n  // Apply sampling if needed\n  const sampledPoints = stratifiedSample(rawPoints, maxSamples);\n\n  const n = sampledPoints.length;\n  const ptsX: number[] = new Array(n);\n  const ptsY: number[] = new Array(n);\n\n  for (let i = 0; i < n; i++) {\n    ptsX[i] = sampledPoints[i].x;\n    ptsY[i] = sampledPoints[i].y;\n  }\n\n  /**\n   * Optimized predictSingle using binary search on pre-sorted data\n   */\n  function predictSingle(x0: number, robustWeights?: number[]) {\n    if (n === 0) {\n      return 0;\n    }\n\n    // Binary search to find insertion point\n    let left = 0;\n    let right = n;\n    while (left < right) {\n      const mid = (left + right) >> 1;\n      if (ptsX[mid] < x0) {\n        left = mid + 1;\n      } else {\n        right = mid;\n      }\n    }\n\n    const m = Math.min(n, Math.max(2, Math.floor(span * n)));\n\n    // Expand by distance: equal numbers on either side need not be the nearest points.\n    let lo = left - 1;\n    let hi = left;\n    for (let count = 0; count < m; count++) {\n      if (lo >= 0 && (hi >= n || x0 - ptsX[lo] <= ptsX[hi] - x0)) {\n        lo--;\n      } else {\n        hi++;\n      }\n    }\n    const actualStart = lo + 1;\n    const end = hi;\n\n    // Find max distance and compute weights in single pass\n    let maxDist = 0;\n    const windowSize = end - actualStart;\n    const distances: number[] = new Array(windowSize);\n\n    for (let i = actualStart; i < end; i++) {\n      const dist = Math.abs(ptsX[i] - x0);\n      distances[i - actualStart] = dist;\n      if (dist > maxDist) {\n        maxDist = dist;\n      }\n    }\n\n    // Compute weights using pre-calculated distances\n    let sumw = 0;\n    const w: number[] = new Array(windowSize);\n    for (let i = 0; i < windowSize; i++) {\n      const u = maxDist === 0 ? 0 : distances[i] / maxDist;\n      let wi = tricube(u);\n      if (robustWeights && robustWeights[actualStart + i] != null) {\n        wi *= robustWeights[actualStart + i];\n      }\n      w[i] = wi;\n      sumw += wi;\n    }\n\n    if (sumw === 0) {\n      // Fall back to the nearest y, preserving input order for equal distances.\n      let nearestIdx = 0;\n      let nearestDistance = Math.abs(ptsX[0] - x0);\n      for (let i = 1; i < n; i++) {\n        const distance = Math.abs(ptsX[i] - x0);\n        if (\n          distance < nearestDistance ||\n          (distance === nearestDistance && sampledPoints[i].index < sampledPoints[nearestIdx].index)\n        ) {\n          nearestIdx = i;\n          nearestDistance = distance;\n        }\n      }\n      return ptsY[nearestIdx];\n    }\n\n    if (degree === 0) {\n      let s = 0;\n      for (let i = 0; i < w.length; i++) {\n        s += w[i] * ptsY[actualStart + i];\n      }\n      return s / sumw;\n    }\n\n    // weighted linear regression on local points\n    let sw = 0;\n    let sx = 0;\n    let sy = 0;\n    let sxx = 0;\n    let sxy = 0;\n    for (let i = actualStart; i < end; i++) {\n      const idx = i - actualStart;\n      const xi = ptsX[i];\n      const yi = ptsY[i];\n      const wi = w[idx];\n      sw += wi;\n      sx += wi * xi;\n      sy += wi * yi;\n      sxx += wi * xi * xi;\n      sxy += wi * xi * yi;\n    }\n\n    const meanX = sx / sw;\n    const meanY = sy / sw;\n    const denom = sxx - sx * meanX;\n    const slope = Math.abs(denom) < 1e-12 ? 0 : (sxy - sx * meanY) / denom;\n    const intercept = meanY - slope * meanX;\n    return intercept + slope * x0;\n  }\n\n  function predict(x0: number | number[]) {\n    if (Array.isArray(x0)) {\n      const len = x0.length;\n      const out: number[] = new Array(len);\n      for (let i = 0; i < len; i++) {\n        out[i] = predictSingle(x0[i]);\n      }\n      return out;\n    }\n    return predictSingle(x0 as number);\n  }\n\n  function evaluateGrid(N: number) {\n    if (N <= 0) {\n      return [];\n    }\n    if (n === 0) {\n      return [];\n    }\n\n    const out: { x: number; y: number }[] = new Array(N);\n    const min = ptsX[0];\n    const max = ptsX[n - 1];\n\n    if (min === max) {\n      const v = predictSingle(min);\n      for (let i = 0; i < N; i++) {\n        out[i] = { x: min, y: v };\n      }\n      return out;\n    }\n    const step = (max - min) / (N - 1);\n\n    // optionally add robust iterations\n    let robustWeights: number[] | undefined;\n\n    // Disable robustness iterations for large datasets to improve performance\n    // Users can override by setting iterations explicitly\n    const effectiveIterations = options.iterations != null ? iterations : n > 500 ? 0 : iterations;\n\n    if (effectiveIterations > 0) {\n      for (let iter = 0; iter < effectiveIterations; iter++) {\n        // compute fits - pre-allocate arrays\n        const fits: number[] = new Array(n);\n        const res: number[] = new Array(n);\n\n        for (let i = 0; i < n; i++) {\n          fits[i] = predictSingle(ptsX[i], robustWeights);\n          res[i] = Math.abs(ptsY[i] - fits[i]);\n        }\n\n        // median absolute deviation\n        const sortedRes = res.slice().sort((a, b) => a - b);\n        const med = sortedRes[Math.floor(n / 2)] || 0;\n        robustWeights = new Array(n);\n        for (let i = 0; i < n; i++) {\n          const u = med === 0 ? 0 : res[i] / (6 * med);\n          const w = Math.abs(u) >= 1 ? 0 : (1 - u * u) * (1 - u * u);\n          robustWeights[i] = w;\n        }\n      }\n    }\n\n    for (let i = 0; i < N; i++) {\n      const px = i === N - 1 ? max : min + step * i;\n      out[i] = { x: px, y: predictSingle(px, robustWeights) };\n    }\n    return out;\n  }\n\n  function confidenceInterval(N: number = 50) {\n    const out: { x: number; mean: number; lower: number; upper: number; predLower: number; predUpper: number }[] = [];\n\n    if (N <= 0) {\n      return out;\n    }\n    if (n === 0) {\n      return out;\n    }\n\n    // use data x-range (already sorted)\n    const min = ptsX[0];\n    const max = ptsX[n - 1];\n\n    if (min === undefined || max === undefined || min === Infinity || max === -Infinity) {\n      return out;\n    }\n\n    const comps = computeLinearCIComponents(data, x, y, (xx: number) => predictSingle(xx));\n    if (comps.n === 0) {\n      return out;\n    }\n\n    const z = confidenceCriticalValue(alpha);\n    if (comps.min === comps.max) {\n      const v = predictSingle(comps.min);\n      const errs = stdErrorsAt(comps.min, comps);\n      const meanHalfWidth = confidenceHalfWidth(z, errs.seMean);\n      const predHalfWidth = confidenceHalfWidth(z, errs.sePred);\n      for (let i = 0; i < N; i++) {\n        out.push({\n          x: comps.min,\n          mean: v,\n          lower: v - meanHalfWidth,\n          upper: v + meanHalfWidth,\n          predLower: v - predHalfWidth,\n          predUpper: v + predHalfWidth\n        });\n      }\n      return out;\n    }\n\n    const step = (max - min) / (N - 1);\n    for (let i = 0; i < N; i++) {\n      const px = i === N - 1 ? max : min + step * i;\n      const yh = predictSingle(px);\n      const errs = stdErrorsAt(px, comps);\n      const meanHalfWidth = confidenceHalfWidth(z, errs.seMean);\n      const predHalfWidth = confidenceHalfWidth(z, errs.sePred);\n      out.push({\n        x: px,\n        mean: yh,\n        lower: yh - meanHalfWidth,\n        upper: yh + meanHalfWidth,\n        predLower: yh - predHalfWidth,\n        predUpper: yh + predHalfWidth\n      });\n    }\n    return out;\n  }\n\n  return {\n    predict,\n    evaluate: predict as any,\n    evaluateGrid,\n    confidenceInterval\n  };\n}\n\nexport default regressionLowess;\n"]}
+{"version":3,"sources":["../src/common/regression-lowess.ts"],"names":[],"mappings":";;;AAAA,2DAA6G;AAC7G,qCAAkC;AAClC,yDAK4B;AAE5B;;;;GAIG;AACH,SAAgB,gBAAgB,CAC9B,IAAW,EACX,IAAwB,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,EAChC,IAAwB,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,EAChC,UAOI,EAAE;;IAEN,MAAM,IAAI,GAAG,MAAA,OAAO,CAAC,IAAI,mCAAI,CAAC,GAAG,CAAC,CAAC;IACnC,MAAM,MAAM,GAAG,MAAA,OAAO,CAAC,MAAM,mCAAI,CAAC,CAAC;IACnC,MAAM,UAAU,GAAG,MAAA,OAAO,CAAC,UAAU,mCAAI,CAAC,CAAC;IAC3C,MAAM,KAAK,GAAG,MAAA,OAAO,CAAC,KAAK,mCAAI,CAAC,CAAC;IACjC,MAAM,KAAK,GAAG,MAAA,OAAO,CAAC,KAAK,mCAAI,IAAI,CAAC;IACpC,IACE,CAAC,CAAC,IAAI,GAAG,CAAC,IAAI,IAAI,IAAI,CAAC,CAAC;QACxB,CAAC,MAAM,CAAC,SAAS,CAAC,UAAU,CAAC;QAC7B,UAAU,GAAG,CAAC;QACd,CAAC,MAAM,CAAC,QAAQ,CAAC,KAAK,CAAC;QACvB,KAAK,GAAG,CAAC;QACT,CAAC,MAAM,KAAK,CAAC,IAAI,MAAM,KAAK,CAAC,CAAC;QAC9B,CAAC,OAAO,CAAC,UAAU,IAAI,IAAI,IAAI,CAAC,CAAC,MAAM,CAAC,SAAS,CAAC,OAAO,CAAC,UAAU,CAAC,IAAI,OAAO,CAAC,UAAU,GAAG,CAAC,CAAC,CAAC,EACjG;QACA,MAAM,IAAI,UAAU,CAAC,8DAA8D,CAAC,CAAC;KACtF;IACD,IAAI,GAAG,GAAG,IAAA,oCAAgB,EAAC,IAAI,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;IAC9E,IAAI,OAAO,CAAC,UAAU,IAAI,IAAI,IAAI,GAAG,CAAC,MAAM,GAAG,OAAO,CAAC,UAAU,EAAE;QACjE,MAAM,IAAI,GAAG,GAAG,CAAC;QACjB,MAAM,KAAK,GAAG,OAAO,CAAC,UAAU,CAAC;QACjC,GAAG,GAAG,KAAK,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,KAAK,EAAE,EAAE,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,GAAG,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,GAAG,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;KACxG;IACD,MAAM,CAAC,GAAG,GAAG,CAAC,MAAM,CAAC;IACrB,MAAM,IAAI,GAAG,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IAC/B,MAAM,IAAI,GAAG,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IAC/B,MAAM,KAAK,GAAG,IAAA,mCAAe,EAAC,GAAG,CAAC,CAAC;IACnC,MAAM,WAAW,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,IAAA,uCAAmB,EAAC,CAAC,EAAE,KAAK,CAAC,OAAO,EAAE,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC;IACvF,MAAM,WAAW,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,IAAA,uCAAmB,EAAC,CAAC,EAAE,KAAK,CAAC,OAAO,EAAE,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC;IACvF,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,IAAI,GAAG,CAAC,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;IACrE,IAAI,aAAmC,CAAC;IAExC,SAAS,KAAK,CAAC,EAAU;QACvB,IAAI,CAAC,CAAC,EAAE;YACN,OAAO,CAAC,CAAC;SACV;QACD,MAAM,KAAK,GAAG,IAAA,uCAAmB,EAAC,EAAE,EAAE,KAAK,CAAC,OAAO,EAAE,KAAK,CAAC,MAAM,CAAC,CAAC;QACnE,wCAAwC;QACxC,IAAI,IAAI,GAAG,CAAC,CAAC;QACb,IAAI,KAAK,GAAG,CAAC,GAAG,KAAK,CAAC;QACtB,OAAO,IAAI,GAAG,KAAK,EAAE;YACnB,MAAM,GAAG,GAAG,CAAC,IAAI,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC;YACjC,MAAM,aAAa,GAAG,EAAE,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC;YACrC,MAAM,aAAa,GAAG,IAAI,CAAC,GAAG,GAAG,KAAK,CAAC,GAAG,EAAE,CAAC;YAC7C,IACE,MAAM,CAAC,QAAQ,CAAC,aAAa,CAAC,IAAI,MAAM,CAAC,QAAQ,CAAC,aAAa,CAAC;gBAC9D,CAAC,CAAC,aAAa,GAAG,aAAa;gBAC/B,CAAC,CAAC,KAAK,GAAG,WAAW,CAAC,GAAG,CAAC,GAAG,WAAW,CAAC,GAAG,GAAG,KAAK,CAAC,GAAG,KAAK,EAC/D;gBACA,IAAI,GAAG,GAAG,GAAG,CAAC,CAAC;aAChB;iBAAM;gBACL,KAAK,GAAG,GAAG,CAAC;aACb;SACF;QACD,IAAI,KAAK,GAAG,IAAI,CAAC;QACjB,IAAI,GAAG,GAAG,IAAI,GAAG,KAAK,CAAC;QACvB,IAAI,IAAI,CAAC,KAAK,CAAC,KAAK,IAAI,CAAC,GAAG,GAAG,CAAC,CAAC,EAAE;YACjC,OAAO,KAAK,GAAG,CAAC,IAAI,IAAI,CAAC,KAAK,GAAG,CAAC,CAAC,KAAK,IAAI,CAAC,KAAK,CAAC,EAAE;gBACnD,KAAK,EAAE,CAAC;aACT;YACD,OAAO,GAAG,GAAG,CAAC,IAAI,IAAI,CAAC,GAAG,CAAC,KAAK,IAAI,CAAC,KAAK,CAAC,EAAE;gBAC3C,GAAG,EAAE,CAAC;aACP;SACF;QACD,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC,EAAE,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,GAAG,CAAC,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC;QACrF,MAAM,MAAM,GAAG,CAAC,MAAM,CAAC,QAAQ,CAAC,SAAS,CAAC,CAAC;QAC3C,MAAM,MAAM,GAAG,MAAM;YACnB,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,WAAW,CAAC,KAAK,CAAC,GAAG,KAAK,CAAC,EAAE,IAAI,CAAC,GAAG,CAAC,WAAW,CAAC,GAAG,GAAG,CAAC,CAAC,GAAG,KAAK,CAAC,CAAC;YACxF,CAAC,CAAC,SAAS,CAAC;QACd,6BAA6B;QAC7B,IAAI,IAAI,GAAG,QAAQ,CAAC;QACpB,IAAI,IAAI,GAAG,CAAC,QAAQ,CAAC;QACrB,KAAK,IAAI,CAAC,GAAG,KAAK,EAAE,CAAC,GAAG,GAAG,EAAE,CAAC,EAAE,EAAE;YAChC,IAAI,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;YAC/B,IAAI,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;SAChC;QACD,MAAM,OAAO,GAAG,IAAI,GAAG,CAAC,GAAG,IAAI,GAAG,CAAC,CAAC;QACpC,MAAM,MAAM,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,GAAG,OAAO,CAAC,EAAE,IAAI,CAAC,GAAG,CAAC,IAAI,GAAG,OAAO,CAAC,CAAC,IAAI,CAAC,CAAC;QACjF,IAAI,EAAE,GAAG,CAAC,CAAC;QACX,IAAI,EAAE,GAAG,CAAC,CAAC;QACX,IAAI,EAAE,GAAG,CAAC,CAAC;QACX,IAAI,GAAG,GAAG,CAAC,CAAC;QACZ,IAAI,GAAG,GAAG,CAAC,CAAC;QACZ,KAAK,IAAI,CAAC,GAAG,KAAK,EAAE,CAAC,GAAG,GAAG,EAAE,CAAC,EAAE,EAAE;YAChC,MAAM,CAAC,GAAG,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,WAAW,CAAC,CAAC,CAAC,GAAG,KAAK,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC,GAAG,MAAM,CAAC;YACvF,MAAM,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;YACtB,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;YACrC,MAAM,CAAC,GAAG,CAAC,GAAG,CAAC,GAAG,CAAC,GAAG,CAAC,aAAa,CAAC,CAAC,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YAC7D,EAAE,IAAI,CAAC,CAAC;YACR,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC;YACZ,MAAM,QAAQ,GAAG,IAAA,uCAAmB,EAAC,IAAI,CAAC,CAAC,CAAC,EAAE,OAAO,EAAE,MAAM,CAAC,CAAC;YAC/D,EAAE,IAAI,CAAC,GAAG,QAAQ,CAAC;YACnB,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;YACjB,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,QAAQ,CAAC;SACzB;QACD,IAAI,EAAE,KAAK,CAAC,EAAE;YACZ,+BAA+B;YAC/B,IAAI,QAAQ,GAAG,QAAQ,CAAC;YACxB,IAAI,MAAM,GAAG,CAAC,CAAC;YACf,IAAI,KAAK,GAAG,CAAC,CAAC;YACd,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE;gBAC1B,MAAM,CAAC,GAAG,aAAa,CAAC,CAAC,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;gBAC/C,IAAI,CAAC,KAAK,CAAC,EAAE;oBACX,SAAS;iBACV;gBACD,MAAM,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,WAAW,CAAC,CAAC,CAAC,GAAG,KAAK,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC,CAAC;gBACnE,IAAI,CAAC,GAAG,QAAQ,EAAE;oBAChB,QAAQ,GAAG,CAAC,CAAC;oBACb,MAAM,GAAG,CAAC,CAAC;oBACX,KAAK,GAAG,IAAI,CAAC,CAAC,CAAC,CAAC;iBACjB;qBAAM,IAAI,CAAC,KAAK,QAAQ,EAAE;oBACzB,KAAK,GAAG,KAAK,GAAG,CAAC,MAAM,GAAG,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,GAAG,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC;oBACvE,MAAM,IAAI,CAAC,CAAC;iBACb;aACF;YACD,IAAI,MAAM,KAAK,CAAC,EAAE;gBAChB,MAAM,IAAI,UAAU,CAAC,4CAA4C,CAAC,CAAC;aACpE;YACD,OAAO,KAAK,CAAC;SACd;QACD,MAAM,EAAE,GAAG,EAAE,GAAG,EAAE,CAAC;QACnB,MAAM,EAAE,GAAG,EAAE,GAAG,EAAE,CAAC;QACnB,MAAM,QAAQ,GAAG,GAAG,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,CAAC;QACpC,MAAM,KAAK,GAAG,MAAM,KAAK,CAAC,IAAI,QAAQ,IAAI,MAAM,CAAC,OAAO,GAAG,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,CAAC,GAAG,QAAQ,CAAC;QACpG,OAAO,OAAO,GAAG,MAAM,GAAG,CAAC,EAAE,GAAG,KAAK,GAAG,EAAE,CAAC,CAAC;IAC9C,CAAC;IAED,qCAAqC;IACrC,MAAM,OAAO,GAAa,EAAE,CAAC;IAC7B,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,GAAI;QACvB,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;QAChB,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE;YACf,MAAM;SACP;QACD,IAAI,IAAI,GAAG,CAAC,GAAG,CAAC,CAAC;QACjB,IAAI,KAAK,GAAG,CAAC,EAAE;YACb,OAAO,IAAI,GAAG,CAAC,GAAG,CAAC,IAAI,IAAI,CAAC,IAAI,GAAG,CAAC,CAAC,GAAG,IAAI,CAAC,CAAC,CAAC,IAAI,KAAK,EAAE;gBACxD,IAAI,EAAE,CAAC;aACR;SACF;QACD,wBAAwB;QACxB,OAAO,IAAI,GAAG,CAAC,GAAG,CAAC,IAAI,IAAI,CAAC,IAAI,GAAG,CAAC,CAAC,KAAK,IAAI,CAAC,IAAI,CAAC,EAAE;YACpD,IAAI,EAAE,CAAC;SACR;QACD,CAAC,GAAG,IAAI,CAAC;KACV;IACD,IAAI,IAAI,GAAG,IAAI,KAAK,CAAS,CAAC,CAAC,CAAC;IAChC,SAAS,eAAe;QACtB,MAAM,GAAG,GAAG,IAAI,KAAK,CAAS,CAAC,CAAC,CAAC;QACjC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;YACvC,MAAM,KAAK,GAAG,OAAO,CAAC,CAAC,CAAC,CAAC;YACzB,GAAG,CAAC,KAAK,CAAC,GAAG,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC;YAChC,IAAI,CAAC,KAAK,CAAC,EAAE;gBACX,SAAS;aACV;YACD,MAAM,QAAQ,GAAG,OAAO,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;YAChC,MAAM,QAAQ,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,IAAI,CAAC,QAAQ,CAAC,CAAC;YAC9C,MAAM,KAAK,GAAG,MAAM,CAAC,QAAQ,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,WAAW,CAAC,KAAK,CAAC,GAAG,WAAW,CAAC,QAAQ,CAAC,CAAC;YAChG,KAAK,IAAI,CAAC,GAAG,QAAQ,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,EAAE,CAAC,EAAE,EAAE;gBACzC,MAAM,CAAC,GACL,KAAK,KAAK,CAAC;oBACT,CAAC,CAAC,CAAC;oBACH,CAAC,CAAC,CAAC,MAAM,CAAC,QAAQ,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,IAAI,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,WAAW,CAAC,CAAC,CAAC,GAAG,WAAW,CAAC,QAAQ,CAAC,CAAC,GAAG,KAAK,CAAC;gBAC9G,GAAG,CAAC,CAAC,CAAC,GAAG,GAAG,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;aACnD;SACF;QACD,OAAO,GAAG,CAAC;IACb,CAAC;IACD,KAAK,IAAI,SAAS,GAAG,CAAC,EAAE,SAAS,IAAI,UAAU,EAAE,SAAS,EAAE,EAAE;QAC5D,IAAI,GAAG,eAAe,EAAE,CAAC;QACzB,IAAI,SAAS,KAAK,UAAU,EAAE;YAC5B,MAAM;SACP;QACD,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE;YAClC,MAAM,QAAQ,GAAG,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC;YAC7B,OAAO,MAAM,CAAC,QAAQ,CAAC,QAAQ,CAAC;gBAC9B,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,QAAQ,GAAG,KAAK,CAAC,MAAM,CAAC;gBACnC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,WAAW,CAAC,CAAC,CAAC,GAAG,IAAA,uCAAmB,EAAC,CAAC,EAAE,KAAK,CAAC,OAAO,EAAE,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC;QACrF,CAAC,CAAC,CAAC;QACH,MAAM,GAAG,GAAG,IAAA,eAAM,EAAC,SAAS,CAAC,KAAK,EAAE,CAAC,CAAC;QACtC,MAAM,WAAW,GAAG,IAAI,CAAC;QACzB,aAAa,GAAG,SAAS,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE;YACrC,IAAI,GAAG,KAAK,CAAC,EAAE;gBACb,MAAM,SAAS,GAAG,MAAM,CAAC,OAAO,GAAG,EAAE,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,EAAE,IAAI,CAAC,GAAG,CAAC,WAAW,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,KAAK,CAAC,MAAM,CAAC,CAAC;gBAC/G,OAAO,CAAC,IAAI,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;aAC/B;YACD,MAAM,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,GAAG,GAAG,CAAC,CAAC;YACxB,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC;QACvC,CAAC,CAAC,CAAC;KACJ;IACD,SAAS,aAAa,CAAC,EAAU;QAC/B,IAAI,KAAK,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,GAAG,IAAI,CAAC,CAAC,CAAC,IAAI,EAAE,GAAG,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,EAAE;YAC5D,OAAO,KAAK,CAAC,EAAE,CAAC,CAAC;SAClB;QACD,IAAI,EAAE,GAAG,CAAC,CAAC;QACX,IAAI,EAAE,GAAG,CAAC,GAAG,CAAC,CAAC;QACf,OAAO,EAAE,GAAG,EAAE,GAAG,CAAC,EAAE;YAClB,MAAM,GAAG,GAAG,CAAC,EAAE,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC;YAC5B,IAAI,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,EAAE;gBACnB,EAAE,GAAG,GAAG,CAAC;aACV;iBAAM;gBACL,EAAE,GAAG,GAAG,CAAC;aACV;SACF;QACD,IAAI,IAAI,CAAC,EAAE,CAAC,KAAK,IAAI,CAAC,EAAE,CAAC,EAAE;YACzB,OAAO,IAAI,CAAC,EAAE,CAAC,CAAC;SACjB;QACD,MAAM,KAAK,GAAG,IAAI,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,EAAE,CAAC,CAAC;QAClC,MAAM,CAAC,GAAG,MAAM,CAAC,QAAQ,CAAC,KAAK,CAAC;YAC9B,CAAC,CAAC,CAAC,EAAE,GAAG,IAAI,CAAC,EAAE,CAAC,CAAC,GAAG,KAAK;YACzB,CAAC,CAAC,CAAC,IAAA,uCAAmB,EAAC,EAAE,EAAE,KAAK,CAAC,OAAO,EAAE,KAAK,CAAC,MAAM,CAAC,GAAG,WAAW,CAAC,EAAE,CAAC,CAAC,GAAG,CAAC,WAAW,CAAC,EAAE,CAAC,GAAG,WAAW,CAAC,EAAE,CAAC,CAAC,CAAC;QACnH,OAAO,IAAI,CAAC,EAAE,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,IAAI,CAAC,EAAE,CAAC,GAAG,CAAC,CAAC;IAC3C,CAAC;IACD,SAAS,OAAO,CAAC,EAAqB;QACpC,OAAO,KAAK,CAAC,OAAO,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,GAAG,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC,aAAa,CAAC,EAAE,CAAC,CAAC;IACvE,CAAC;IACD,SAAS,YAAY,CAAC,CAAS;QAC7B,OAAO,IAAA,kCAAc,EAAC,IAAI,CAAC,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,EAAE,CAAC,EAAE,aAAa,CAAC,CAAC;IAChE,CAAC;IAED,SAAS,kBAAkB,CAAC,IAAY,EAAE;QACxC,MAAM,GAAG,GAAsG,EAAE,CAAC;QAElH,IAAI,CAAC,IAAI,CAAC,EAAE;YACV,OAAO,GAAG,CAAC;SACZ;QACD,IAAI,CAAC,KAAK,CAAC,EAAE;YACX,OAAO,GAAG,CAAC;SACZ;QAED,oCAAoC;QACpC,MAAM,GAAG,GAAG,IAAI,CAAC,CAAC,CAAC,CAAC;QACpB,MAAM,GAAG,GAAG,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;QAExB,IAAI,GAAG,KAAK,SAAS,IAAI,GAAG,KAAK,SAAS,IAAI,GAAG,KAAK,QAAQ,IAAI,GAAG,KAAK,CAAC,QAAQ,EAAE;YACnF,OAAO,GAAG,CAAC;SACZ;QAED,MAAM,KAAK,GAAG,IAAA,4CAAyB,EAAC,IAAI,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAU,EAAE,EAAE,CAAC,aAAa,CAAC,EAAE,CAAC,CAAC,CAAC;QACvF,IAAI,KAAK,CAAC,CAAC,KAAK,CAAC,EAAE;YACjB,OAAO,GAAG,CAAC;SACZ;QAED,MAAM,CAAC,GAAG,IAAA,0CAAuB,EAAC,KAAK,CAAC,CAAC;QACzC,IAAI,KAAK,CAAC,GAAG,KAAK,KAAK,CAAC,GAAG,EAAE;YAC3B,MAAM,CAAC,GAAG,aAAa,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;YACnC,MAAM,IAAI,GAAG,IAAA,8BAAW,EAAC,KAAK,CAAC,GAAG,EAAE,KAAK,CAAC,CAAC;YAC3C,MAAM,aAAa,GAAG,IAAA,sCAAmB,EAAC,CAAC,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;YAC1D,MAAM,aAAa,GAAG,IAAA,sCAAmB,EAAC,CAAC,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;YAC1D,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE;gBAC1B,GAAG,CAAC,IAAI,CAAC;oBACP,CAAC,EAAE,KAAK,CAAC,GAAG;oBACZ,IAAI,EAAE,CAAC;oBACP,KAAK,EAAE,CAAC,GAAG,aAAa;oBACxB,KAAK,EAAE,CAAC,GAAG,aAAa;oBACxB,SAAS,EAAE,CAAC,GAAG,aAAa;oBAC5B,SAAS,EAAE,CAAC,GAAG,aAAa;iBAC7B,CAAC,CAAC;aACJ;YACD,OAAO,GAAG,CAAC;SACZ;QAED,MAAM,IAAI,GAAG,CAAC,GAAG,GAAG,GAAG,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;QACnC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE;YAC1B,MAAM,EAAE,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,GAAG,IAAI,GAAG,CAAC,CAAC;YAC9C,MAAM,EAAE,GAAG,aAAa,CAAC,EAAE,CAAC,CAAC;YAC7B,MAAM,IAAI,GAAG,IAAA,8BAAW,EAAC,EAAE,EAAE,KAAK,CAAC,CAAC;YACpC,MAAM,aAAa,GAAG,IAAA,sCAAmB,EAAC,CAAC,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;YAC1D,MAAM,aAAa,GAAG,IAAA,sCAAmB,EAAC,CAAC,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;YAC1D,GAAG,CAAC,IAAI,CAAC;gBACP,CAAC,EAAE,EAAE;gBACL,IAAI,EAAE,EAAE;gBACR,KAAK,EAAE,EAAE,GAAG,aAAa;gBACzB,KAAK,EAAE,EAAE,GAAG,aAAa;gBACzB,SAAS,EAAE,EAAE,GAAG,aAAa;gBAC7B,SAAS,EAAE,EAAE,GAAG,aAAa;aAC9B,CAAC,CAAC;SACJ;QACD,OAAO,GAAG,CAAC;IACb,CAAC;IAED,OAAO;QACL,OAAO;QACP,QAAQ,EAAE,OAAO;QACjB,YAAY;QACZ,kBAAkB;QAClB,0BAA0B;QAC1B,MAAM,EAAE,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC;KACpD,CAAC;AACJ,CAAC;AA7SD,4CA6SC;AACD,kBAAe,gBAAgB,CAAC","file":"regression-lowess.js","sourcesContent":["import { regressionPoints, regressionGrid, regressionBasis, normalizeRegression } from './regression-solver';\nimport { median } from './median';\nimport {\n  computeLinearCIComponents,\n  confidenceCriticalValue,\n  confidenceHalfWidth,\n  stdErrorsAt\n} from './regression-utils';\n\n/**\n * 局部线性 LOWESS。默认全量观测、span=2/3、三次稳健重加权。\n * delta 显式插值加速；maxSamples 仅保留显式选择的历史抽样用途。\n * predict 与 evaluateGrid 查询构建时固定的同一模型，不触发再拟合。\n */\nexport function regressionLowess(\n  data: any[],\n  x: (d: any) => number = d => d.x,\n  y: (d: any) => number = d => d.y,\n  options: {\n    span?: number;\n    degree?: 1 | 0;\n    iterations?: number;\n    alpha?: number;\n    maxSamples?: number;\n    delta?: number;\n  } = {}\n) {\n  const span = options.span ?? 2 / 3;\n  const degree = options.degree ?? 1;\n  const iterations = options.iterations ?? 3;\n  const delta = options.delta ?? 0;\n  const alpha = options.alpha ?? 0.05;\n  if (\n    !(span > 0 && span <= 1) ||\n    !Number.isInteger(iterations) ||\n    iterations < 0 ||\n    !Number.isFinite(delta) ||\n    delta < 0 ||\n    (degree !== 0 && degree !== 1) ||\n    (options.maxSamples != null && (!Number.isInteger(options.maxSamples) || options.maxSamples < 2))\n  ) {\n    throw new RangeError('Invalid LOWESS span, degree, iterations, delta or maxSamples');\n  }\n  let pts = regressionPoints(data, x, y).sort((a, b) => a.x - b.x || a.y - b.y);\n  if (options.maxSamples != null && pts.length > options.maxSamples) {\n    const full = pts;\n    const count = options.maxSamples;\n    pts = Array.from({ length: count }, (_, i) => full[Math.floor((i * (full.length - 1)) / (count - 1))]);\n  }\n  const n = pts.length;\n  const ptsX = pts.map(p => p.x);\n  const ptsY = pts.map(p => p.y);\n  const basis = regressionBasis(pts);\n  const normalizedX = ptsX.map(v => normalizeRegression(v, basis.xCenter, basis.xScale));\n  const normalizedY = ptsY.map(v => normalizeRegression(v, basis.yCenter, basis.yScale));\n  const count = Math.min(n, Math.max(2, Math.floor(span * n + 1e-10)));\n  let robustWeights: number[] | undefined;\n\n  function exact(x0: number) {\n    if (!n) {\n      return 0;\n    }\n    const query = normalizeRegression(x0, basis.xCenter, basis.xScale);\n    // 找到连续的最近邻窗口；相等距离选择较小的 x（ties 的 y 已排序）。\n    let left = 0;\n    let right = n - count;\n    while (left < right) {\n      const mid = (left + right) >>> 1;\n      const lowerDistance = x0 - ptsX[mid];\n      const upperDistance = ptsX[mid + count] - x0;\n      if (\n        Number.isFinite(lowerDistance) && Number.isFinite(upperDistance)\n          ? lowerDistance > upperDistance\n          : query - normalizedX[mid] > normalizedX[mid + count] - query\n      ) {\n        left = mid + 1;\n      } else {\n        right = mid;\n      }\n    }\n    let start = left;\n    let end = left + count;\n    if (ptsX[start] === ptsX[end - 1]) {\n      while (start > 0 && ptsX[start - 1] === ptsX[start]) {\n        start--;\n      }\n      while (end < n && ptsX[end] === ptsX[start]) {\n        end++;\n      }\n    }\n    const rawRadius = Math.max(Math.abs(ptsX[start] - x0), Math.abs(ptsX[end - 1] - x0));\n    const scaled = !Number.isFinite(rawRadius);\n    const radius = scaled\n      ? Math.max(Math.abs(normalizedX[start] - query), Math.abs(normalizedX[end - 1] - query))\n      : rawRadius;\n    // 局部响应坐标不能由远处、未参与该邻域的观测决定精度。\n    let yMin = Infinity;\n    let yMax = -Infinity;\n    for (let i = start; i < end; i++) {\n      yMin = Math.min(yMin, ptsY[i]);\n      yMax = Math.max(yMax, ptsY[i]);\n    }\n    const yCenter = yMin / 2 + yMax / 2;\n    const yScale = Math.max(Math.abs(yMin - yCenter), Math.abs(yMax - yCenter)) || 1;\n    let sw = 0;\n    let sx = 0;\n    let sy = 0;\n    let sxx = 0;\n    let sxy = 0;\n    for (let i = start; i < end; i++) {\n      const z = radius === 0 ? 0 : (scaled ? normalizedX[i] - query : ptsX[i] - x0) / radius;\n      const u = Math.abs(z);\n      const t = u >= 1 ? 0 : 1 - u * u * u;\n      const w = t * t * t * (robustWeights ? robustWeights[i] : 1);\n      sw += w;\n      sx += w * z;\n      const response = normalizeRegression(ptsY[i], yCenter, yScale);\n      sy += w * response;\n      sxx += w * z * z;\n      sxy += w * z * response;\n    }\n    if (sw === 0) {\n      // 核边界零权重时使用最近有效观测的加权常数；等距离取均值。\n      let distance = Infinity;\n      let weight = 0;\n      let value = 0;\n      for (let i = 0; i < n; i++) {\n        const w = robustWeights ? robustWeights[i] : 1;\n        if (w === 0) {\n          continue;\n        }\n        const d = Math.abs(scaled ? normalizedX[i] - query : ptsX[i] - x0);\n        if (d < distance) {\n          distance = d;\n          weight = w;\n          value = ptsY[i];\n        } else if (d === distance) {\n          value = value * (weight / (weight + w)) + ptsY[i] * (w / (weight + w));\n          weight += w;\n        }\n      }\n      if (weight === 0) {\n        throw new RangeError('LOWESS has no effective local observations');\n      }\n      return value;\n    }\n    const mx = sx / sw;\n    const my = sy / sw;\n    const variance = sxx / sw - mx * mx;\n    const slope = degree === 0 || variance <= Number.EPSILON * 32 ? 0 : (sxy / sw - mx * my) / variance;\n    return yCenter + yScale * (my - slope * mx);\n  }\n\n  // delta 不删除观测。只减少局部回归求值位置，全部残差参与重加权。\n  const anchors: number[] = [];\n  for (let i = 0; i < n; ) {\n    anchors.push(i);\n    if (i === n - 1) {\n      break;\n    }\n    let next = i + 1;\n    if (delta > 0) {\n      while (next + 1 < n && ptsX[next + 1] - ptsX[i] <= delta) {\n        next++;\n      }\n    }\n    // 重复 x 一起使用同一预测；端点不能丢失。\n    while (next + 1 < n && ptsX[next + 1] === ptsX[next]) {\n      next++;\n    }\n    i = next;\n  }\n  let fits = new Array<number>(n);\n  function fitObservations() {\n    const out = new Array<number>(n);\n    for (let a = 0; a < anchors.length; a++) {\n      const index = anchors[a];\n      out[index] = exact(ptsX[index]);\n      if (a === 0) {\n        continue;\n      }\n      const previous = anchors[a - 1];\n      const rawWidth = ptsX[index] - ptsX[previous];\n      const width = Number.isFinite(rawWidth) ? rawWidth : normalizedX[index] - normalizedX[previous];\n      for (let i = previous + 1; i < index; i++) {\n        const t =\n          width === 0\n            ? 0\n            : (Number.isFinite(rawWidth) ? ptsX[i] - ptsX[previous] : normalizedX[i] - normalizedX[previous]) / width;\n        out[i] = out[previous] * (1 - t) + out[index] * t;\n      }\n    }\n    return out;\n  }\n  for (let iteration = 0; iteration <= iterations; iteration++) {\n    fits = fitObservations();\n    if (iteration === iterations) {\n      break;\n    }\n    const residuals = fits.map((v, i) => {\n      const residual = ptsY[i] - v;\n      return Number.isFinite(residual)\n        ? Math.abs(residual / basis.yScale)\n        : Math.abs(normalizedY[i] - normalizeRegression(v, basis.yCenter, basis.yScale));\n    });\n    const med = median(residuals.slice());\n    const currentFits = fits;\n    robustWeights = residuals.map((r, i) => {\n      if (med === 0) {\n        const tolerance = Number.EPSILON * 64 * (Math.max(Math.abs(ptsY[i]), Math.abs(currentFits[i])) / basis.yScale);\n        return r <= tolerance ? 1 : 0;\n      }\n      const u = r / (6 * med);\n      return u >= 1 ? 0 : (1 - u * u) ** 2;\n    });\n  }\n  function predictSingle(x0: number): number {\n    if (delta === 0 || n < 2 || x0 < ptsX[0] || x0 > ptsX[n - 1]) {\n      return exact(x0);\n    }\n    let lo = 0;\n    let hi = n - 1;\n    while (hi - lo > 1) {\n      const mid = (lo + hi) >>> 1;\n      if (ptsX[mid] <= x0) {\n        lo = mid;\n      } else {\n        hi = mid;\n      }\n    }\n    if (ptsX[hi] === ptsX[lo]) {\n      return fits[lo];\n    }\n    const width = ptsX[hi] - ptsX[lo];\n    const t = Number.isFinite(width)\n      ? (x0 - ptsX[lo]) / width\n      : (normalizeRegression(x0, basis.xCenter, basis.xScale) - normalizedX[lo]) / (normalizedX[hi] - normalizedX[lo]);\n    return fits[lo] * (1 - t) + fits[hi] * t;\n  }\n  function predict(x0: number | number[]) {\n    return Array.isArray(x0) ? x0.map(predictSingle) : predictSingle(x0);\n  }\n  function evaluateGrid(N: number) {\n    return regressionGrid(ptsX[0], ptsX[n - 1], N, predictSingle);\n  }\n\n  function confidenceInterval(N: number = 50) {\n    const out: { x: number; mean: number; lower: number; upper: number; predLower: number; predUpper: number }[] = [];\n\n    if (N <= 0) {\n      return out;\n    }\n    if (n === 0) {\n      return out;\n    }\n\n    // use data x-range (already sorted)\n    const min = ptsX[0];\n    const max = ptsX[n - 1];\n\n    if (min === undefined || max === undefined || min === Infinity || max === -Infinity) {\n      return out;\n    }\n\n    const comps = computeLinearCIComponents(data, x, y, (xx: number) => predictSingle(xx));\n    if (comps.n === 0) {\n      return out;\n    }\n\n    const z = confidenceCriticalValue(alpha);\n    if (comps.min === comps.max) {\n      const v = predictSingle(comps.min);\n      const errs = stdErrorsAt(comps.min, comps);\n      const meanHalfWidth = confidenceHalfWidth(z, errs.seMean);\n      const predHalfWidth = confidenceHalfWidth(z, errs.sePred);\n      for (let i = 0; i < N; i++) {\n        out.push({\n          x: comps.min,\n          mean: v,\n          lower: v - meanHalfWidth,\n          upper: v + meanHalfWidth,\n          predLower: v - predHalfWidth,\n          predUpper: v + predHalfWidth\n        });\n      }\n      return out;\n    }\n\n    const step = (max - min) / (N - 1);\n    for (let i = 0; i < N; i++) {\n      const px = i === N - 1 ? max : min + step * i;\n      const yh = predictSingle(px);\n      const errs = stdErrorsAt(px, comps);\n      const meanHalfWidth = confidenceHalfWidth(z, errs.seMean);\n      const predHalfWidth = confidenceHalfWidth(z, errs.sePred);\n      out.push({\n        x: px,\n        mean: yh,\n        lower: yh - meanHalfWidth,\n        upper: yh + meanHalfWidth,\n        predLower: yh - predHalfWidth,\n        predUpper: yh + predHalfWidth\n      });\n    }\n    return out;\n  }\n\n  return {\n    predict,\n    evaluate: predict,\n    evaluateGrid,\n    confidenceInterval,\n    // 节点值按 x 排序，重复观测仍保持其统计权重。\n    fitted: pts.map((p, i) => ({ x: p.x, y: fits[i] }))\n  };\n}\nexport default regressionLowess;\n"]}

package/cjs/common/regression-polynomial.d.ts

@@ -3,6 +3,7 @@ export declare function regressionPolynomial(data: any[], x?: (d: any) => number
     alpha?: number;
 }): {
     degree: number;
+    rank: number;
     coef: number[];
     predict: (xx: number) => number;
     rSquared: number;

package/cjs/common/regression-polynomial.js

@@ -1,82 +1,21 @@
 "use strict";
 
-var __importDefault = this && this.__importDefault || function(mod) {
-    return mod && mod.__esModule ? mod : {
-        default: mod
-    };
-};
-
 Object.defineProperty(exports, "__esModule", {
     value: !0
 }), exports.regressionPolynomial = void 0;
 
-const regression_linear_1 = require("./regression-linear"), isNil_1 = __importDefault(require("./isNil")), regression_utils_1 = require("./regression-utils");
-
-function solveLinearSystem(A, b) {
-    // Gaussian elimination with partial pivoting
-    const n = b.length, M = new Array(n);
-    // clone
-        for (let i = 0; i < n; i++) M[i] = A[i].slice(), M[i].push(b[i]);
-    for (let k = 0; k < n; k++) {
-        // find pivot
-        let maxRow = k, maxVal = Math.abs(M[k][k]);
-        for (let i = k + 1; i < n; i++) {
-            const v = Math.abs(M[i][k]);
-            v > maxVal && (maxVal = v, maxRow = i);
-        }
-        if (maxRow !== k) {
-            const tmp = M[k];
-            M[k] = M[maxRow], M[maxRow] = tmp;
-        }
-        // singular check
-                if (Math.abs(M[k][k]) < 1e-12) {
-            return new Array(n).fill(0);
-        }
-        // normalize row
-                for (let j = k + 1; j <= n; j++) M[k][j] = M[k][j] / M[k][k];
-        M[k][k] = 1;
-        // eliminate
-        for (let i = 0; i < n; i++) {
-            if (i === k) continue;
-            const factor = M[i][k];
-            if (0 !== factor) {
-                for (let j = k + 1; j <= n; j++) M[i][j] -= factor * M[k][j];
-                M[i][k] = 0;
-            }
-        }
-    }
-    const x = new Array(n);
-    for (let i = 0; i < n; i++) x[i] = M[i][n];
-    return x;
-}
+const regression_linear_1 = require("./regression-linear"), regression_solver_1 = require("./regression-solver"), regression_utils_1 = require("./regression-utils");
 
 function regressionPolynomial(data, x = (d => d.x), y = (d => d.y), options = {}) {
     var _a, _b;
-    let degree = null !== (_a = options.degree) && void 0 !== _a ? _a : 0;
-    degree < 0 && (degree = 0);
-    const alpha = null !== (_b = options.alpha) && void 0 !== _b ? _b : .05, m = degree + 1, sums = new Array(2 * degree + 1).fill(0);
-    (0, regression_linear_1.visitPoints)(data, x, y, ((dx, dy) => {
-        let xp = 1;
-        for (let k = 0; k < sums.length; k++) sums[k] += xp, xp *= dx;
-    }));
-    // build normal matrix
-    const A = new Array(m);
-    for (let i = 0; i < m; i++) {
-        A[i] = new Array(m).fill(0);
-        for (let j = 0; j < m; j++) A[i][j] = sums[i + j];
-    }
-    const B = new Array(m).fill(0);
-    (0, regression_linear_1.visitPoints)(data, x, y, ((dx, dy) => {
-        let xp = 1;
-        for (let k = 0; k < m; k++) B[k] += dy * xp, xp *= dx;
-    }));
-    const coef = solveLinearSystem(A, B), predict = xx => {
-        let xp = 1, v = 0;
-        for (let k = 0; k < coef.length; k++) v += coef[k] * xp, xp *= xx;
-        return v;
-    };
+    const degree = null !== (_a = options.degree) && void 0 !== _a ? _a : 0;
+    if (!Number.isInteger(degree) || degree < 0) throw new RangeError("Polynomial degree must be an nonnegative integer");
+    const alpha = null !== (_b = options.alpha) && void 0 !== _b ? _b : .05, points = (0, 
+    regression_solver_1.regressionPoints)(data, x, y), model = (0, regression_solver_1.fitRegression)(points, degree), coef = (0, 
+    regression_solver_1.regressionCoefficients)(model), predict = xx => (0, regression_solver_1.regressionPrediction)(model, xx);
     return {
         degree: degree,
+        rank: model.rank,
         coef: coef,
         predict: predict,
         rSquared: (0, regression_linear_1.rSquared)(data, x, y, (() => {
@@ -87,30 +26,10 @@ function regressionPolynomial(data, x = (d => d.x), y = (d => d.y), options = {}
             })), 0 === cnt ? 0 : sum / cnt;
         })(), predict),
         evaluateGrid(N) {
-            const out = [];
-            if (N <= 0) return out;
-            // compute range
-                        let min = 1 / 0, max = -1 / 0;
-            if ((0, regression_linear_1.visitPoints)(data, x, y, (dx => {
-                (0, isNil_1.default)(dx) || (dx < min && (min = dx), dx > max && (max = dx));
-            })), min === 1 / 0 || max === -1 / 0) return out;
-            if (min === max) {
-                const v = predict(min);
-                for (let i = 0; i < N; i++) out.push({
-                    x: min,
-                    y: v
-                });
-                return out;
-            }
-            const step = (max - min) / (N - 1);
-            for (let i = 0; i < N; i++) {
-                const px = i === N - 1 ? max : min + step * i;
-                out.push({
-                    x: px,
-                    y: predict(px)
-                });
-            }
-            return out;
+            let min = 1 / 0, max = -1 / 0;
+            return points.forEach((p => {
+                min = Math.min(min, p.x), max = Math.max(max, p.x);
+            })), (0, regression_solver_1.regressionGrid)(min, max, N, predict);
         },
         confidenceInterval(N = 50) {
             const out = [];

package/cjs/common/regression-polynomial.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../src/common/regression-polynomial.ts"],"names":[],"mappings":";;;;;;AAAA,2DAA4D;AAC5D,oDAA4B;AAC5B,yDAK4B;AAE5B,SAAS,iBAAiB,CAAC,CAAa,EAAE,CAAW;IACnD,6CAA6C;IAC7C,MAAM,CAAC,GAAG,CAAC,CAAC,MAAM,CAAC;IACnB,QAAQ;IACR,MAAM,CAAC,GAAe,IAAI,KAAK,CAAC,CAAC,CAAC,CAAC;IACnC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE;QAC1B,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,KAAK,EAAE,CAAC;QACpB,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;KACjB;IAED,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE;QAC1B,aAAa;QACb,IAAI,MAAM,GAAG,CAAC,CAAC;QACf,IAAI,MAAM,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QAC/B,KAAK,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE;YAC9B,MAAM,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YAC5B,IAAI,CAAC,GAAG,MAAM,EAAE;gBACd,MAAM,GAAG,CAAC,CAAC;gBACX,MAAM,GAAG,CAAC,CAAC;aACZ;SACF;QACD,IAAI,MAAM,KAAK,CAAC,EAAE;YAChB,MAAM,GAAG,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;YACjB,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,MAAM,CAAC,CAAC;YACjB,CAAC,CAAC,MAAM,CAAC,GAAG,GAAG,CAAC;SACjB;QAED,iBAAiB;QACjB,IAAI,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,KAAK,EAAE;YAC7B,sCAAsC;YACtC,MAAM,GAAG,GAAa,IAAI,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;YAC3C,OAAO,GAAG,CAAC;SACZ;QAED,gBAAgB;QAChB,KAAK,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC,IAAI,CAAC,EAAE,CAAC,EAAE,EAAE;YAC/B,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;SAC7B;QACD,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC;QAEZ,YAAY;QACZ,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE;YAC1B,IAAI,CAAC,KAAK,CAAC,EAAE;gBACX,SAAS;aACV;YACD,MAAM,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YACvB,IAAI,MAAM,KAAK,CAAC,EAAE;gBAChB,SAAS;aACV;YACD,KAAK,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC,IAAI,CAAC,EAAE,CAAC,EAAE,EAAE;gBAC/B,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;aAC7B;YACD,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC;SACb;KACF;IAED,MAAM,CAAC,GAAa,IAAI,KAAK,CAAC,CAAC,CAAC,CAAC;IACjC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE;QAC1B,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;KAChB;IACD,OAAO,CAAC,CAAC;AACX,CAAC;AAED,SAAgB,oBAAoB,CAClC,IAAW,EACX,IAAwB,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,EAChC,IAAwB,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,EAChC,UAA+C,EAAE;;IAEjD,IAAI,MAAM,GAAG,MAAA,OAAO,CAAC,MAAM,mCAAI,CAAC,CAAC;IACjC,IAAI,MAAM,GAAG,CAAC,EAAE;QACd,MAAM,GAAG,CAAC,CAAC;KACZ;IACD,MAAM,KAAK,GAAG,MAAA,OAAO,CAAC,KAAK,mCAAI,IAAI,CAAC;IACpC,MAAM,CAAC,GAAG,MAAM,GAAG,CAAC,CAAC;IACrB,MAAM,IAAI,GAAa,IAAI,KAAK,CAAC,CAAC,GAAG,MAAM,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;IAEzD,IAAA,+BAAW,EAAC,IAAI,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,EAAE,EAAE,EAAE,EAAE;QACjC,IAAI,EAAE,GAAG,CAAC,CAAC;QACX,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;YACpC,IAAI,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;YACd,EAAE,IAAI,EAAE,CAAC;SACV;IACH,CAAC,CAAC,CAAC;IAEH,sBAAsB;IACtB,MAAM,CAAC,GAAe,IAAI,KAAK,CAAC,CAAC,CAAC,CAAC;IACnC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE;QAC1B,CAAC,CAAC,CAAC,CAAC,GAAG,IAAI,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;QAC5B,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE;YAC1B,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;SACvB;KACF;IAED,MAAM,CAAC,GAAa,IAAI,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;IACzC,IAAA,+BAAW,EAAC,IAAI,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,EAAE,EAAE,EAAE,EAAE;QACjC,IAAI,EAAE,GAAG,CAAC,CAAC;QACX,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE;YAC1B,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,GAAG,EAAE,CAAC;YAChB,EAAE,IAAI,EAAE,CAAC;SACV;IACH,CAAC,CAAC,CAAC;IAEH,MAAM,IAAI,GAAG,iBAAiB,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;IAErC,MAAM,OAAO,GAAG,CAAC,EAAU,EAAE,EAAE;QAC7B,IAAI,EAAE,GAAG,CAAC,CAAC;QACX,IAAI,CAAC,GAAG,CAAC,CAAC;QACV,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;YACpC,CAAC,IAAI,IAAI,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC;YAClB,EAAE,IAAI,EAAE,CAAC;SACV;QACD,OAAO,CAAC,CAAC;IACX,CAAC,CAAC;IAEF,OAAO;QACL,MAAM;QACN,IAAI;QACJ,OAAO;QACP,QAAQ,EAAE,IAAA,4BAAQ,EAChB,IAAI,EACJ,CAAC,EACD,CAAC,EACD,CAAC,GAAG,EAAE;YACJ,iBAAiB;YACjB,IAAI,GAAG,GAAG,CAAC,CAAC;YACZ,IAAI,GAAG,GAAG,CAAC,CAAC;YACZ,IAAA,+BAAW,EAAC,IAAI,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,GAAG,EAAE,EAAE,EAAE,EAAE;gBAClC,GAAG,IAAI,EAAE,CAAC;gBACV,GAAG,EAAE,CAAC;YACR,CAAC,CAAC,CAAC;YACH,OAAO,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,GAAG,GAAG,CAAC;QACnC,CAAC,CAAC,EAAE,EACJ,OAAO,CACR;QACD,YAAY,CAAC,CAAS;YACpB,MAAM,GAAG,GAA+B,EAAE,CAAC;YAC3C,IAAI,CAAC,IAAI,CAAC,EAAE;gBACV,OAAO,GAAG,CAAC;aACZ;YACD,gBAAgB;YAChB,IAAI,GAAG,GAAG,QAAQ,CAAC;YACnB,IAAI,GAAG,GAAG,CAAC,QAAQ,CAAC;YACpB,IAAA,+BAAW,EAAC,IAAI,EAAE,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,EAAE;gBAC3B,IAAI,CAAC,IAAA,eAAK,EAAC,EAAE,CAAC,EAAE;oBACd,IAAI,EAAE,GAAG,GAAG,EAAE;wBACZ,GAAG,GAAG,EAAE,CAAC;qBACV;oBACD,IAAI,EAAE,GAAG,GAAG,EAAE;wBACZ,GAAG,GAAG,EAAE,CAAC;qBACV;iBACF;YACH,CAAC,CAAC,CAAC;YACH,IAAI,GAAG,KAAK,QAAQ,IAAI,GAAG,KAAK,CAAC,QAAQ,EAAE;gBACzC,OAAO,GAAG,CAAC;aACZ;YACD,IAAI,GAAG,KAAK,GAAG,EAAE;gBACf,MAAM,CAAC,GAAG,OAAO,CAAC,GAAG,CAAC,CAAC;gBACvB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE;oBAC1B,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC;iBAC5B;gBACD,OAAO,GAAG,CAAC;aACZ;YACD,MAAM,IAAI,GAAG,CAAC,GAAG,GAAG,GAAG,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;YACnC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE;gBAC1B,MAAM,EAAE,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,GAAG,IAAI,GAAG,CAAC,CAAC;gBAC9C,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,OAAO,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC;aACrC;YACD,OAAO,GAAG,CAAC;QACb,CAAC;QACD,kBAAkB,CAAC,IAAY,EAAE;YAC/B,MAAM,GAAG,GAAsG,EAAE,CAAC;YAElH,IAAI,CAAC,IAAI,CAAC,EAAE;gBACV,OAAO,GAAG,CAAC;aACZ;YAED,MAAM,KAAK,GAAG,IAAA,4CAAyB,EAAC,IAAI,EAAE,CAAC,EAAE,CAAC,EAAE,OAAO,CAAC,CAAC;YAC7D,IAAI,KAAK,CAAC,CAAC,KAAK,CAAC,EAAE;gBACjB,OAAO,GAAG,CAAC;aACZ;YAED,MAAM,CAAC,GAAG,IAAA,0CAAuB,EAAC,KAAK,CAAC,CAAC;YACzC,IAAI,KAAK,CAAC,GAAG,KAAK,KAAK,CAAC,GAAG,EAAE;gBAC3B,MAAM,CAAC,GAAG,OAAO,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;gBAC7B,MAAM,IAAI,GAAG,IAAA,8BAAW,EAAC,KAAK,CAAC,GAAG,EAAE,KAAK,CAAC,CAAC;gBAC3C,MAAM,aAAa,GAAG,IAAA,sCAAmB,EAAC,CAAC,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;gBAC1D,MAAM,aAAa,GAAG,IAAA,sCAAmB,EAAC,CAAC,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;gBAC1D,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE;oBAC1B,GAAG,CAAC,IAAI,CAAC;wBACP,CAAC,EAAE,KAAK,CAAC,GAAG;wBACZ,IAAI,EAAE,CAAC;wBACP,KAAK,EAAE,CAAC,GAAG,aAAa;wBACxB,KAAK,EAAE,CAAC,GAAG,aAAa;wBACxB,SAAS,EAAE,CAAC,GAAG,aAAa;wBAC5B,SAAS,EAAE,CAAC,GAAG,aAAa;qBAC7B,CAAC,CAAC;iBACJ;gBACD,OAAO,GAAG,CAAC;aACZ;YAED,MAAM,IAAI,GAAG,CAAC,KAAK,CAAC,GAAG,GAAG,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;YAC/C,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE;gBAC1B,MAAM,EAAE,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,KAAK,CAAC,GAAG,GAAG,IAAI,GAAG,CAAC,CAAC;gBAC1D,MAAM,EAAE,GAAG,OAAO,CAAC,EAAE,CAAC,CAAC;gBACvB,MAAM,IAAI,GAAG,IAAA,8BAAW,EAAC,EAAE,EAAE,KAAK,CAAC,CAAC;gBACpC,MAAM,aAAa,GAAG,IAAA,sCAAmB,EAAC,CAAC,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;gBAC1D,MAAM,aAAa,GAAG,IAAA,sCAAmB,EAAC,CAAC,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;gBAC1D,GAAG,CAAC,IAAI,CAAC;oBACP,CAAC,EAAE,EAAE;oBACL,IAAI,EAAE,EAAE;oBACR,KAAK,EAAE,EAAE,GAAG,aAAa;oBACzB,KAAK,EAAE,EAAE,GAAG,aAAa;oBACzB,SAAS,EAAE,EAAE,GAAG,aAAa;oBAC7B,SAAS,EAAE,EAAE,GAAG,aAAa;iBAC9B,CAAC,CAAC;aACJ;YACD,OAAO,GAAG,CAAC;QACb,CAAC;KACF,CAAC;AACJ,CAAC;AA7JD,oDA6JC;AAED,kBAAe,oBAAoB,CAAC","file":"regression-polynomial.js","sourcesContent":["import { visitPoints, rSquared } from './regression-linear';\nimport isNil from './isNil';\nimport {\n  computeLinearCIComponents,\n  confidenceCriticalValue,\n  confidenceHalfWidth,\n  stdErrorsAt\n} from './regression-utils';\n\nfunction solveLinearSystem(A: number[][], b: number[]): number[] {\n  // Gaussian elimination with partial pivoting\n  const n = b.length;\n  // clone\n  const M: number[][] = new Array(n);\n  for (let i = 0; i < n; i++) {\n    M[i] = A[i].slice();\n    M[i].push(b[i]);\n  }\n\n  for (let k = 0; k < n; k++) {\n    // find pivot\n    let maxRow = k;\n    let maxVal = Math.abs(M[k][k]);\n    for (let i = k + 1; i < n; i++) {\n      const v = Math.abs(M[i][k]);\n      if (v > maxVal) {\n        maxVal = v;\n        maxRow = i;\n      }\n    }\n    if (maxRow !== k) {\n      const tmp = M[k];\n      M[k] = M[maxRow];\n      M[maxRow] = tmp;\n    }\n\n    // singular check\n    if (Math.abs(M[k][k]) < 1e-12) {\n      // return least squares fallback zeros\n      const res: number[] = new Array(n).fill(0);\n      return res;\n    }\n\n    // normalize row\n    for (let j = k + 1; j <= n; j++) {\n      M[k][j] = M[k][j] / M[k][k];\n    }\n    M[k][k] = 1;\n\n    // eliminate\n    for (let i = 0; i < n; i++) {\n      if (i === k) {\n        continue;\n      }\n      const factor = M[i][k];\n      if (factor === 0) {\n        continue;\n      }\n      for (let j = k + 1; j <= n; j++) {\n        M[i][j] -= factor * M[k][j];\n      }\n      M[i][k] = 0;\n    }\n  }\n\n  const x: number[] = new Array(n);\n  for (let i = 0; i < n; i++) {\n    x[i] = M[i][n];\n  }\n  return x;\n}\n\nexport function regressionPolynomial(\n  data: any[],\n  x: (d: any) => number = d => d.x,\n  y: (d: any) => number = d => d.y,\n  options: { degree?: number; alpha?: number } = {}\n) {\n  let degree = options.degree ?? 0;\n  if (degree < 0) {\n    degree = 0;\n  }\n  const alpha = options.alpha ?? 0.05;\n  const m = degree + 1;\n  const sums: number[] = new Array(2 * degree + 1).fill(0);\n\n  visitPoints(data, x, y, (dx, dy) => {\n    let xp = 1;\n    for (let k = 0; k < sums.length; k++) {\n      sums[k] += xp;\n      xp *= dx;\n    }\n  });\n\n  // build normal matrix\n  const A: number[][] = new Array(m);\n  for (let i = 0; i < m; i++) {\n    A[i] = new Array(m).fill(0);\n    for (let j = 0; j < m; j++) {\n      A[i][j] = sums[i + j];\n    }\n  }\n\n  const B: number[] = new Array(m).fill(0);\n  visitPoints(data, x, y, (dx, dy) => {\n    let xp = 1;\n    for (let k = 0; k < m; k++) {\n      B[k] += dy * xp;\n      xp *= dx;\n    }\n  });\n\n  const coef = solveLinearSystem(A, B);\n\n  const predict = (xx: number) => {\n    let xp = 1;\n    let v = 0;\n    for (let k = 0; k < coef.length; k++) {\n      v += coef[k] * xp;\n      xp *= xx;\n    }\n    return v;\n  };\n\n  return {\n    degree,\n    coef,\n    predict,\n    rSquared: rSquared(\n      data,\n      x,\n      y,\n      (() => {\n        // compute mean y\n        let sum = 0;\n        let cnt = 0;\n        visitPoints(data, x, y, (_dx, dy) => {\n          sum += dy;\n          cnt++;\n        });\n        return cnt === 0 ? 0 : sum / cnt;\n      })(),\n      predict\n    ),\n    evaluateGrid(N: number) {\n      const out: { x: number; y: number }[] = [];\n      if (N <= 0) {\n        return out;\n      }\n      // compute range\n      let min = Infinity;\n      let max = -Infinity;\n      visitPoints(data, x, y, dx => {\n        if (!isNil(dx)) {\n          if (dx < min) {\n            min = dx;\n          }\n          if (dx > max) {\n            max = dx;\n          }\n        }\n      });\n      if (min === Infinity || max === -Infinity) {\n        return out;\n      }\n      if (min === max) {\n        const v = predict(min);\n        for (let i = 0; i < N; i++) {\n          out.push({ x: min, y: v });\n        }\n        return out;\n      }\n      const step = (max - min) / (N - 1);\n      for (let i = 0; i < N; i++) {\n        const px = i === N - 1 ? max : min + step * i;\n        out.push({ x: px, y: predict(px) });\n      }\n      return out;\n    },\n    confidenceInterval(N: number = 50) {\n      const out: { x: number; mean: number; lower: number; upper: number; predLower: number; predUpper: number }[] = [];\n\n      if (N <= 0) {\n        return out;\n      }\n\n      const comps = computeLinearCIComponents(data, x, y, predict);\n      if (comps.n === 0) {\n        return out;\n      }\n\n      const z = confidenceCriticalValue(alpha);\n      if (comps.min === comps.max) {\n        const v = predict(comps.min);\n        const errs = stdErrorsAt(comps.min, comps);\n        const meanHalfWidth = confidenceHalfWidth(z, errs.seMean);\n        const predHalfWidth = confidenceHalfWidth(z, errs.sePred);\n        for (let i = 0; i < N; i++) {\n          out.push({\n            x: comps.min,\n            mean: v,\n            lower: v - meanHalfWidth,\n            upper: v + meanHalfWidth,\n            predLower: v - predHalfWidth,\n            predUpper: v + predHalfWidth\n          });\n        }\n        return out;\n      }\n\n      const step = (comps.max - comps.min) / (N - 1);\n      for (let i = 0; i < N; i++) {\n        const px = i === N - 1 ? comps.max : comps.min + step * i;\n        const yh = predict(px);\n        const errs = stdErrorsAt(px, comps);\n        const meanHalfWidth = confidenceHalfWidth(z, errs.seMean);\n        const predHalfWidth = confidenceHalfWidth(z, errs.sePred);\n        out.push({\n          x: px,\n          mean: yh,\n          lower: yh - meanHalfWidth,\n          upper: yh + meanHalfWidth,\n          predLower: yh - predHalfWidth,\n          predUpper: yh + predHalfWidth\n        });\n      }\n      return out;\n    }\n  };\n}\n\nexport default regressionPolynomial;\n"]}
+{"version":3,"sources":["../src/common/regression-polynomial.ts"],"names":[],"mappings":";;;AAAA,2DAA4D;AAC5D,2DAM6B;AAC7B,yDAK4B;AAE5B,SAAgB,oBAAoB,CAClC,IAAW,EACX,IAAwB,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,EAChC,IAAwB,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,EAChC,UAA+C,EAAE;;IAEjD,MAAM,MAAM,GAAG,MAAA,OAAO,CAAC,MAAM,mCAAI,CAAC,CAAC;IACnC,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,MAAM,CAAC,IAAI,MAAM,GAAG,CAAC,EAAE;QAC3C,MAAM,IAAI,UAAU,CAAC,kDAAkD,CAAC,CAAC;KAC1E;IACD,MAAM,KAAK,GAAG,MAAA,OAAO,CAAC,KAAK,mCAAI,IAAI,CAAC;IACpC,MAAM,MAAM,GAAG,IAAA,oCAAgB,EAAC,IAAI,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC;IAC5C,MAAM,KAAK,GAAG,IAAA,iCAAa,EAAC,MAAM,EAAE,MAAM,CAAC,CAAC;IAC5C,MAAM,IAAI,GAAG,IAAA,0CAAsB,EAAC,KAAK,CAAC,CAAC;IAC3C,MAAM,OAAO,GAAG,CAAC,EAAU,EAAE,EAAE,CAAC,IAAA,wCAAoB,EAAC,KAAK,EAAE,EAAE,CAAC,CAAC;IAEhE,OAAO;QACL,MAAM;QACN,IAAI,EAAE,KAAK,CAAC,IAAI;QAChB,IAAI;QACJ,OAAO;QACP,QAAQ,EAAE,IAAA,4BAAQ,EAChB,IAAI,EACJ,CAAC,EACD,CAAC,EACD,CAAC,GAAG,EAAE;YACJ,iBAAiB;YACjB,IAAI,GAAG,GAAG,CAAC,CAAC;YACZ,IAAI,GAAG,GAAG,CAAC,CAAC;YACZ,IAAA,+BAAW,EAAC,IAAI,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,GAAG,EAAE,EAAE,EAAE,EAAE;gBAClC,GAAG,IAAI,EAAE,CAAC;gBACV,GAAG,EAAE,CAAC;YACR,CAAC,CAAC,CAAC;YACH,OAAO,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,GAAG,GAAG,CAAC;QACnC,CAAC,CAAC,EAAE,EACJ,OAAO,CACR;QACD,YAAY,CAAC,CAAS;YACpB,IAAI,GAAG,GAAG,QAAQ,CAAC;YACnB,IAAI,GAAG,GAAG,CAAC,QAAQ,CAAC;YACpB,MAAM,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE;gBACjB,GAAG,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC;gBACzB,GAAG,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC;YAC3B,CAAC,CAAC,CAAC;YACH,OAAO,IAAA,kCAAc,EAAC,GAAG,EAAE,GAAG,EAAE,CAAC,EAAE,OAAO,CAAC,CAAC;QAC9C,CAAC;QACD,kBAAkB,CAAC,IAAY,EAAE;YAC/B,MAAM,GAAG,GAAsG,EAAE,CAAC;YAElH,IAAI,CAAC,IAAI,CAAC,EAAE;gBACV,OAAO,GAAG,CAAC;aACZ;YAED,MAAM,KAAK,GAAG,IAAA,4CAAyB,EAAC,IAAI,EAAE,CAAC,EAAE,CAAC,EAAE,OAAO,CAAC,CAAC;YAC7D,IAAI,KAAK,CAAC,CAAC,KAAK,CAAC,EAAE;gBACjB,OAAO,GAAG,CAAC;aACZ;YAED,MAAM,CAAC,GAAG,IAAA,0CAAuB,EAAC,KAAK,CAAC,CAAC;YACzC,IAAI,KAAK,CAAC,GAAG,KAAK,KAAK,CAAC,GAAG,EAAE;gBAC3B,MAAM,CAAC,GAAG,OAAO,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;gBAC7B,MAAM,IAAI,GAAG,IAAA,8BAAW,EAAC,KAAK,CAAC,GAAG,EAAE,KAAK,CAAC,CAAC;gBAC3C,MAAM,aAAa,GAAG,IAAA,sCAAmB,EAAC,CAAC,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;gBAC1D,MAAM,aAAa,GAAG,IAAA,sCAAmB,EAAC,CAAC,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;gBAC1D,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE;oBAC1B,GAAG,CAAC,IAAI,CAAC;wBACP,CAAC,EAAE,KAAK,CAAC,GAAG;wBACZ,IAAI,EAAE,CAAC;wBACP,KAAK,EAAE,CAAC,GAAG,aAAa;wBACxB,KAAK,EAAE,CAAC,GAAG,aAAa;wBACxB,SAAS,EAAE,CAAC,GAAG,aAAa;wBAC5B,SAAS,EAAE,CAAC,GAAG,aAAa;qBAC7B,CAAC,CAAC;iBACJ;gBACD,OAAO,GAAG,CAAC;aACZ;YAED,MAAM,IAAI,GAAG,CAAC,KAAK,CAAC,GAAG,GAAG,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;YAC/C,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE;gBAC1B,MAAM,EAAE,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,KAAK,CAAC,GAAG,GAAG,IAAI,GAAG,CAAC,CAAC;gBAC1D,MAAM,EAAE,GAAG,OAAO,CAAC,EAAE,CAAC,CAAC;gBACvB,MAAM,IAAI,GAAG,IAAA,8BAAW,EAAC,EAAE,EAAE,KAAK,CAAC,CAAC;gBACpC,MAAM,aAAa,GAAG,IAAA,sCAAmB,EAAC,CAAC,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;gBAC1D,MAAM,aAAa,GAAG,IAAA,sCAAmB,EAAC,CAAC,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;gBAC1D,GAAG,CAAC,IAAI,CAAC;oBACP,CAAC,EAAE,EAAE;oBACL,IAAI,EAAE,EAAE;oBACR,KAAK,EAAE,EAAE,GAAG,aAAa;oBACzB,KAAK,EAAE,EAAE,GAAG,aAAa;oBACzB,SAAS,EAAE,EAAE,GAAG,aAAa;oBAC7B,SAAS,EAAE,EAAE,GAAG,aAAa;iBAC9B,CAAC,CAAC;aACJ;YACD,OAAO,GAAG,CAAC;QACb,CAAC;KACF,CAAC;AACJ,CAAC;AAhGD,oDAgGC;AAED,kBAAe,oBAAoB,CAAC","file":"regression-polynomial.js","sourcesContent":["import { visitPoints, rSquared } from './regression-linear';\nimport {\n  regressionPoints,\n  fitRegression,\n  regressionPrediction,\n  regressionCoefficients,\n  regressionGrid\n} from './regression-solver';\nimport {\n  computeLinearCIComponents,\n  confidenceCriticalValue,\n  confidenceHalfWidth,\n  stdErrorsAt\n} from './regression-utils';\n\nexport function regressionPolynomial(\n  data: any[],\n  x: (d: any) => number = d => d.x,\n  y: (d: any) => number = d => d.y,\n  options: { degree?: number; alpha?: number } = {}\n) {\n  const degree = options.degree ?? 0;\n  if (!Number.isInteger(degree) || degree < 0) {\n    throw new RangeError('Polynomial degree must be an nonnegative integer');\n  }\n  const alpha = options.alpha ?? 0.05;\n  const points = regressionPoints(data, x, y);\n  const model = fitRegression(points, degree);\n  const coef = regressionCoefficients(model);\n  const predict = (xx: number) => regressionPrediction(model, xx);\n\n  return {\n    degree,\n    rank: model.rank,\n    coef,\n    predict,\n    rSquared: rSquared(\n      data,\n      x,\n      y,\n      (() => {\n        // compute mean y\n        let sum = 0;\n        let cnt = 0;\n        visitPoints(data, x, y, (_dx, dy) => {\n          sum += dy;\n          cnt++;\n        });\n        return cnt === 0 ? 0 : sum / cnt;\n      })(),\n      predict\n    ),\n    evaluateGrid(N: number) {\n      let min = Infinity;\n      let max = -Infinity;\n      points.forEach(p => {\n        min = Math.min(min, p.x);\n        max = Math.max(max, p.x);\n      });\n      return regressionGrid(min, max, N, predict);\n    },\n    confidenceInterval(N: number = 50) {\n      const out: { x: number; mean: number; lower: number; upper: number; predLower: number; predUpper: number }[] = [];\n\n      if (N <= 0) {\n        return out;\n      }\n\n      const comps = computeLinearCIComponents(data, x, y, predict);\n      if (comps.n === 0) {\n        return out;\n      }\n\n      const z = confidenceCriticalValue(alpha);\n      if (comps.min === comps.max) {\n        const v = predict(comps.min);\n        const errs = stdErrorsAt(comps.min, comps);\n        const meanHalfWidth = confidenceHalfWidth(z, errs.seMean);\n        const predHalfWidth = confidenceHalfWidth(z, errs.sePred);\n        for (let i = 0; i < N; i++) {\n          out.push({\n            x: comps.min,\n            mean: v,\n            lower: v - meanHalfWidth,\n            upper: v + meanHalfWidth,\n            predLower: v - predHalfWidth,\n            predUpper: v + predHalfWidth\n          });\n        }\n        return out;\n      }\n\n      const step = (comps.max - comps.min) / (N - 1);\n      for (let i = 0; i < N; i++) {\n        const px = i === N - 1 ? comps.max : comps.min + step * i;\n        const yh = predict(px);\n        const errs = stdErrorsAt(px, comps);\n        const meanHalfWidth = confidenceHalfWidth(z, errs.seMean);\n        const predHalfWidth = confidenceHalfWidth(z, errs.sePred);\n        out.push({\n          x: px,\n          mean: yh,\n          lower: yh - meanHalfWidth,\n          upper: yh + meanHalfWidth,\n          predLower: yh - predHalfWidth,\n          predUpper: yh + predHalfWidth\n        });\n      }\n      return out;\n    }\n  };\n}\n\nexport default regressionPolynomial;\n"]}

package/cjs/common/regression-solver.d.ts

@@ -0,0 +1,26 @@
+/** 回归共用的稳定坐标与小型最小二乘求解器；不构造 XᵀX。 */
+export interface RegressionPoint {
+    x: number;
+    y: number;
+}
+export interface RegressionBasis {
+    xCenter: number;
+    xScale: number;
+    yCenter: number;
+    yScale: number;
+}
+export declare function regressionPoints(data: readonly any[], x: (d: any) => number, y: (d: any) => number): RegressionPoint[];
+export declare function normalizeRegression(value: number, center: number, scale: number): number;
+export declare function regressionBasis(points: readonly RegressionPoint[]): RegressionBasis;
+export declare function fitRegression(points: readonly RegressionPoint[], degree: number, basis?: RegressionBasis): {
+    coef: number[];
+    rank: number;
+    basis: RegressionBasis;
+};
+export declare function regressionPrediction(model: ReturnType<typeof fitRegression>, x: number): number;
+/** 保留公开的升幂 coef；预测始终使用稳定基底。 */
+export declare function regressionCoefficients(model: ReturnType<typeof fitRegression>): number[];
+export declare function regressionGrid(min: number, max: number, count: number, predict: (x: number) => number): {
+    x: number;
+    y: number;
+}[];