Linear Regression function implemented

integration_tests/test_statistics.py

@@ -1,10 +1,11 @@
 from statistics import (mean, fmean, geometric_mean, harmonic_mean, variance,
-                        stdev, pvariance, pstdev, correlation, covariance)
+                        stdev, pvariance, pstdev, correlation, covariance, linear_regression)
 from ltypes import i32, f64, i64, f32
 
 eps: f64
 eps = 1e-12
 
+
 def test_mean():
     b: list[i32]
     b = [9, 4, 10]
@@ -163,6 +164,28 @@ def test_correlation():
     j = correlation(c, d)
     assert abs(j - 0.9057925526720572) < eps
 
+def test_linear_regression():
+    c: list[f64]
+    c = [2.74, 1.23, 2.63, 2.22, 3.0, 1.98]
+    d: list[f64]
+    d = [9.4, 1.23, 2.63, 22.4, 1.9, 13.98]
+
+    slope: f64
+    intercept: f64
+    slope, intercept = linear_regression(c, d)
+
+    assert abs(slope + 0.6098133124816717) < eps
+    assert abs(intercept - 9.992570618707845) < eps
+
+    a: list[i32]
+    b: list[i32]
+    a = [12, 24, 2, 1, 43, 53, 23]
+    b = [2, 13, 14, 63, 49, 7, 3]
+
+    slope, intercept = linear_regression(a, b)
+
+    assert abs(slope + 0.18514007308160782) < eps
+    assert abs(intercept - 25.750304506699152) < eps
 
 def check():
     test_mean()
@@ -173,6 +196,7 @@ def check():
     test_stdev()
     test_pvariance()
     test_pstdev()
+    test_linear_regression()
     test_correlation()
     test_covariance()
 

src/runtime/statistics.py

@@ -322,7 +322,6 @@ def pstdev(x: list[i32]) -> f64:
     """
     return pvariance(x)**0.5
 
-
 @overload
 def correlation(x: list[i32], y: list[i32]) -> f64:
     """
@@ -422,3 +421,80 @@ def covariance(x: list[f64], y: list[f64]) -> f64:
     for i in range(n):
         num += (x[i] - xmean) * (y[i] - ymean)
     return num / (n-1)
+
+@overload
+def linear_regression(x: list[i32], y: list[i32]) -> tuple[f64, f64]:
+
+    """
+    Returns the slope and intercept of simple linear regression
+    parameters estimated using ordinary least squares.
+    """
+    n: i32 = len(x)
+    if len(y) !=  n:
+        raise Exception('linear regression requires that both inputs have same number of data points')
+    if n < 2:
+        raise Exception('linear regression requires at least two data points')
+    xmean: f64 = mean(x)
+    ymean: f64 = mean(y)
+
+    sxy: f64 = 0.0
+    i: i32
+    for i in range(n):
+        sxy += (x[i] - xmean) * (y[i] - ymean)
+
+    sxx: f64 = 0.0
+    j: i32
+    for j in range(n):
+        sxx += (x[j] - xmean) ** 2
+
+    slope: f64
+
+    if sxx == 0:
+        raise Exception('x is constant')
+    else:
+        slope = sxy / sxx
+
+    intercept: f64  = ymean - slope * xmean
+
+    LinReg: tuple[f64, f64] = (slope, intercept)
+
+    return LinReg
+
+@overload
+def linear_regression(x: list[f64], y: list[f64]) -> tuple[f64, f64]:
+
+    """
+    Returns the slope and intercept of simple linear regression
+    parameters estimated using ordinary least squares.
+    """
+    n: i32 = len(x)
+    if len(y) !=  n:
+        raise Exception('linear regression requires that both inputs have same number of data points')
+    if n < 2:
+        raise Exception('linear regression requires at least two data points')
+    xmean: f64 = mean(x)
+    ymean: f64 = mean(y)
+
+    sxy: f64 = 0.0
+    i: i32
+    for i in range(n):
+        sxy += (x[i] - xmean) * (y[i] - ymean)
+
+    sxx: f64 = 0.0
+    j: i32
+    for j in range(n):
+        sxx += (x[j] - xmean) ** 2
+
+    slope: f64
+
+    if sxx == 0:
+        raise Exception('x is constant')
+    else:
+        slope = sxy / sxx
+
+    intercept: f64  = ymean - slope * xmean
+
+    LinReg: tuple[f64, f64] = (slope, intercept)
+
+    return LinReg
+