From 36ba8e95d756670bee76262bd01a5954c4a4364c Mon Sep 17 00:00:00 2001
From: Swayam Bhoir <swayambhoir579@gmail.com>
Date: Sun, 21 Dec 2025 16:46:06 +0530
Subject: [PATCH] Add class oriented linear regression model.

---
 data.csv             |  1 +
 linear_regression.py | 68 ++++++++++++++++++++++++++++++++++++++++++++
 test.py              | 26 +++++++++++++++++
 3 files changed, 95 insertions(+)
 create mode 100644 linear_regression.py
 create mode 100644 test.py

diff --git a/data.csv b/data.csv
index 9e8e26c..2e61e5c 100644
--- a/data.csv
+++ b/data.csv
@@ -1,3 +1,4 @@
+x,y
 32.502345269453031,31.70700584656992
 53.426804033275019,68.77759598163891
 61.530358025636438,62.562382297945803
diff --git a/linear_regression.py b/linear_regression.py
new file mode 100644
index 0000000..5658ba0
--- /dev/null
+++ b/linear_regression.py
@@ -0,0 +1,68 @@
+from typing import List
+
+class LinearRegressionModel:
+    def __init__(self):
+        self.xs = []
+        self.ys = []
+        self.m = 0.0
+        self.b = 0.0
+    
+    def points(self, xs: List[float], ys: List[float]):
+        self.xs = xs
+        self.ys = ys
+
+    def _squared_error(self) -> float:
+        error = 0.0
+        n = len(self.xs)
+
+        for i in range(n):
+            error += (self.ys[i] - (self.m * self.xs[i] + self.b)) ** 2
+        
+        return error
+    
+    def _gradient_step(self):
+        m_gradient = 0.0
+        b_gradient = 0.0
+        n = len(self.xs)
+
+        for i in range(n):
+            m_gradient += -2 * (self.ys[i] - self.m * self.xs[i] - self.b) * self.xs[i]
+            b_gradient += -2 * (self.ys[i] - self.m * self.xs[i] - self.b)
+        
+        return [ m_gradient, b_gradient ]
+
+    def _add_gradient(self, learning_rate: float):
+        m_gradient, b_gradient = self._gradient_step()
+        self.m -= (learning_rate * m_gradient)
+        self.b -= (learning_rate * b_gradient)
+
+    def train(self, epochs: int = 500, learning_rate: float = 0.001):
+        assert len(self.xs) == len(self.ys), "Input and output lists must have the same length."
+
+        assert len(self.xs) > 2, "Data must contain at least 3 points."
+
+        for epoch in range(epochs):
+            self._add_gradient(learning_rate)
+    
+    def predict(self, x: float) -> float:
+        return self.m * x + self.b
+    
+if __name__ == "__main__":
+    import random
+
+    random.seed(42)
+
+    xs = [i for i in range(100)]
+    ys = [2 * x + 3 + random.uniform(-10, 10) for x in xs]
+
+    model = LinearRegressionModel()
+    model.points(xs, ys)
+
+    model.train(epochs=100, learning_rate=0.000001)
+
+    print(f"Trained model: y = {model.m}x + {model.b}")
+
+    # Test prediction
+    
+    test_x = 50
+    print(f"Prediction for x={test_x}: y={model.predict(test_x)}")
diff --git a/test.py b/test.py
new file mode 100644
index 0000000..f144bdc
--- /dev/null
+++ b/test.py
@@ -0,0 +1,26 @@
+import pandas as pd
+import matplotlib.pyplot as plt
+from linear_regression import LinearRegressionModel
+
+points = pd.read_csv("./data.csv")
+
+xs = points['x'].values
+ys  = points['y'].values
+
+model = LinearRegressionModel()
+model.points(xs, ys)
+model.train(epochs=100, learning_rate=0.0000001)
+
+print(f"m = {model.m}, b = {model.b}")
+
+plt.scatter(xs, ys, color='blue', label='Data Points')
+
+predictions = [
+    model.predict(x) for x in xs
+]
+
+plt.plot(xs, predictions, color='red', label='Regression Line')
+plt.xlabel('x')
+plt.ylabel('y')
+plt.legend()
+plt.show()