GH-102670: Use sumprod() to simplify, speed up, and improve accuracy of statistics...

diff --git a/Lib/statistics.py b/Lib/statistics.py
index 7d5d750193a5abdc6c4a69ad9ee6d6b3a262b7c8..6bd214bbfe2ff5976b19408b08d2fcdd9edc66fa 100644 (file)
--- a/Lib/statistics.py
+++ b/Lib/statistics.py
@@ -1036,7 +1036,7 @@ def covariance(x, y, /):
         raise StatisticsError('covariance requires at least two data points')
     xbar = fsum(x) / n
     ybar = fsum(y) / n
-    sxy = fsum((xi - xbar) * (yi - ybar) for xi, yi in zip(x, y))
+    sxy = sumprod((xi - xbar for xi in x), (yi - ybar for yi in y))
     return sxy / (n - 1)
 
 
@@ -1074,11 +1074,14 @@ def correlation(x, y, /, *, method='linear'):
         start = (n - 1) / -2            # Center rankings around zero
         x = _rank(x, start=start)
         y = _rank(y, start=start)
-    xbar = fsum(x) / n
-    ybar = fsum(y) / n
-    sxy = fsum((xi - xbar) * (yi - ybar) for xi, yi in zip(x, y))
-    sxx = fsum((d := xi - xbar) * d for xi in x)
-    syy = fsum((d := yi - ybar) * d for yi in y)
+    else:
+        xbar = fsum(x) / n
+        ybar = fsum(y) / n
+        x = [xi - xbar for xi in x]
+        y = [yi - ybar for yi in y]
+    sxy = sumprod(x, y)
+    sxx = sumprod(x, x)
+    syy = sumprod(y, y)
     try:
         return sxy / sqrt(sxx * syy)
     except ZeroDivisionError:
@@ -1131,14 +1134,13 @@ def linear_regression(x, y, /, *, proportional=False):
         raise StatisticsError('linear regression requires that both inputs have same number of data points')
     if n < 2:
         raise StatisticsError('linear regression requires at least two data points')
-    if proportional:
-        sxy = fsum(xi * yi for xi, yi in zip(x, y))
-        sxx = fsum(xi * xi for xi in x)
-    else:
+    if not proportional:
         xbar = fsum(x) / n
         ybar = fsum(y) / n
-        sxy = fsum((xi - xbar) * (yi - ybar) for xi, yi in zip(x, y))
-        sxx = fsum((d := xi - xbar) * d for xi in x)
+        x = [xi - xbar for xi in x]  # List because used three times below
+        y = (yi - ybar for yi in y)  # Generator because only used once below
+    sxy = sumprod(x, y) + 0.0        # Add zero to coerce result to a float
+    sxx = sumprod(x, x)
     try:
         slope = sxy / sxx   # equivalent to:  covariance(x, y) / variance(x)
     except ZeroDivisionError:

diff --git a/Lib/test/test_statistics.py b/Lib/test/test_statistics.py
index 31a3cb6b53a6f2ab4f56821df9bd936e7b8a57d5..f0fa6454b1f91a70d0548c49b4587c1029126e8c 100644 (file)
--- a/Lib/test/test_statistics.py
+++ b/Lib/test/test_statistics.py
@@ -1,4 +1,4 @@
-"""Test suite for statistics module, including helper NumericTestCase and
+x = """Test suite for statistics module, including helper NumericTestCase and
 approx_equal function.
 
 """
@@ -2610,6 +2610,16 @@ class TestLinearRegression(unittest.TestCase):
         self.assertAlmostEqual(slope, 20 + 1/150)
         self.assertEqual(intercept, 0.0)
 
+    def test_float_output(self):
+        x = [Fraction(2, 3), Fraction(3, 4)]
+        y = [Fraction(4, 5), Fraction(5, 6)]
+        slope, intercept = statistics.linear_regression(x, y)
+        self.assertTrue(isinstance(slope, float))
+        self.assertTrue(isinstance(intercept, float))
+        slope, intercept = statistics.linear_regression(x, y, proportional=True)
+        self.assertTrue(isinstance(slope, float))
+        self.assertTrue(isinstance(intercept, float))
+
 class TestNormalDist:
 
     # General note on precision: The pdf(), cdf(), and overlap() methods

diff --git a/Misc/NEWS.d/next/Library/2023-03-13-18-27-00.gh-issue-102670.GyoThv.rst b/Misc/NEWS.d/next/Library/2023-03-13-18-27-00.gh-issue-102670.GyoThv.rst
new file mode 100644 (file)
index 0000000..3de09f8
--- /dev/null
+++ b/Misc/NEWS.d/next/Library/2023-03-13-18-27-00.gh-issue-102670.GyoThv.rst
@@ -0,0 +1,2 @@
+Optimized fmean(), correlation(), covariance(), and linear_regression()
+using the new math.sumprod() function.