[Autoloop: perf-comparison] Iteration 390: add OLS multiple regression benchmark pair

Run: https://github.com/githubnext/tsb/actions/runs/29003458535

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Author: github-actions[bot]

benchmarks/pandas/bench_ols.py

@@ -0,0 +1,31 @@
+"""Benchmark: OLS (Ordinary Least Squares) multiple regression on 10k rows x 5 predictors"""
+import json, time
+import numpy as np
+
+ROWS = 10_000
+WARMUP = 5
+ITERATIONS = 20
+
+rng = np.random.default_rng(42)
+X = rng.uniform(-1, 1, size=(ROWS, 5))
+# y = 1*x1 + 2*x2 - 0.5*x3 + 3*x4 + 0.1*x5 + noise
+coefs = np.array([1.0, 2.0, -0.5, 3.0, 0.1])
+y = X @ coefs + rng.normal(0, 0.05, size=ROWS)
+
+# Add intercept column (matching tsb OLS default addIntercept=true)
+X_design = np.column_stack([X, np.ones(ROWS)])
+
+for _ in range(WARMUP):
+    np.linalg.lstsq(X_design, y, rcond=None)
+
+start = time.perf_counter()
+for _ in range(ITERATIONS):
+    np.linalg.lstsq(X_design, y, rcond=None)
+total = (time.perf_counter() - start) * 1000
+
+print(json.dumps({
+    "function": "ols",
+    "mean_ms": total / ITERATIONS,
+    "iterations": ITERATIONS,
+    "total_ms": total,
+}))

benchmarks/tsb/bench_ols.ts

@@ -0,0 +1,48 @@
+/**
+ * Benchmark: OLS (Ordinary Least Squares) multiple regression on 10k rows × 5 predictors
+ */
+import { OLS } from "../../src/index.js";
+
+const ROWS = 10_000;
+const WARMUP = 5;
+const ITERATIONS = 20;
+
+// Reproducible design matrix: 5 predictors with known coefficients
+const rng = (seed: number) => {
+  let s = seed;
+  return () => {
+    s = (s * 1664525 + 1013904223) & 0xffffffff;
+    return (s >>> 0) / 4294967296;
+  };
+};
+const rand = rng(42);
+
+const X: number[][] = Array.from({ length: ROWS }, () =>
+  Array.from({ length: 5 }, () => rand() * 2 - 1),
+);
+// y = 1*x1 + 2*x2 - 0.5*x3 + 3*x4 + 0.1*x5 + noise
+const y: number[] = X.map((row) => {
+  const [x1, x2, x3, x4, x5] = row as [number, number, number, number, number];
+  return x1 + 2 * x2 - 0.5 * x3 + 3 * x4 + 0.1 * x5 + (rand() - 0.5) * 0.1;
+});
+
+const model = new OLS();
+
+for (let i = 0; i < WARMUP; i++) {
+  model.fit(X, y);
+}
+
+const start = performance.now();
+for (let i = 0; i < ITERATIONS; i++) {
+  model.fit(X, y);
+}
+const total = performance.now() - start;
+
+console.log(
+  JSON.stringify({
+    function: "ols",
+    mean_ms: total / ITERATIONS,
+    iterations: ITERATIONS,
+    total_ms: total,
+  }),
+);