hamzacodesnow989 commited on
Commit
dc0edc3
·
verified ·
1 Parent(s): ce5338b

Upload linear_regression_from_scratch.py

Browse files
Files changed (1) hide show
  1. linear_regression_from_scratch.py +76 -0
linear_regression_from_scratch.py ADDED
@@ -0,0 +1,76 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import pickle
2
+
3
+
4
+ class LinearRegression:
5
+ def __init__(self, learning_rate=0.01, epochs=1000):
6
+ self.learning_rate = learning_rate
7
+ self.epochs = epochs
8
+ self.w1 = 0.0
9
+ self.w2 = 0.0
10
+ self.b = 0.0
11
+
12
+ def predict(self, x1, x2):
13
+ return self.w1 * x1 + self.w2 * x2 + self.b
14
+
15
+ def compute_gradients(self, X1, X2, y):
16
+ n = len(y)
17
+ dw1, dw2, db = 0.0, 0.0, 0.0
18
+
19
+ for i in range(n):
20
+ y_pred = self.predict(X1[i], X2[i])
21
+ error = y[i] - y_pred
22
+ dw1 += -2 * X1[i] * error
23
+ dw2 += -2 * X2[i] * error
24
+ db += -2 * error
25
+
26
+ return dw1 / n, dw2 / n, db / n
27
+
28
+ def update(self, X1, X2, y):
29
+ dw1, dw2, db = self.compute_gradients(X1, X2, y)
30
+ self.w1 -= self.learning_rate * dw1
31
+ self.w2 -= self.learning_rate * dw2
32
+ self.b -= self.learning_rate * db
33
+
34
+ def mean_squared_error(self, y, y_pred):
35
+ n = len(y)
36
+ return sum((y[i] - y_pred[i]) ** 2 for i in range(n)) / n
37
+
38
+ def fit(self, X1, X2, y):
39
+ for epoch in range(self.epochs):
40
+ self.update(X1, X2, y)
41
+
42
+ if epoch % 200 == 0:
43
+ y_pred = [self.predict(X1[i], X2[i]) for i in range(len(y))]
44
+ mse = self.mean_squared_error(y, y_pred)
45
+ print(f"Epoch {epoch}: MSE = {mse:.4f}")
46
+
47
+
48
+ # -------------------------------
49
+ # TRAINING
50
+ # -------------------------------
51
+ X1 = [2, 4, 6, 8]
52
+ X2 = [2, 4, 6, 8]
53
+ y = [4, 8, 12, 16]
54
+
55
+ model = LinearRegression(learning_rate=0.01, epochs=1000)
56
+ model.fit(X1, X2, y)
57
+
58
+ print("Prediction before saving:", model.predict(2, 2))
59
+
60
+
61
+ # -------------------------------
62
+ # SAVE MODEL
63
+ # -------------------------------
64
+ with open("linear_regression_model.pkl", "wb") as f:
65
+ pickle.dump(model, f)
66
+
67
+ print("Model saved!")
68
+
69
+
70
+ # -------------------------------
71
+ # LOAD MODEL
72
+ # -------------------------------
73
+ with open("linear_regression_model.pkl", "rb") as f:
74
+ loaded_model = pickle.load(f)
75
+
76
+ print("Prediction after loading:", loaded_model.predict(2, 2))