更新 function.py

function.py

@@ -28,7 +28,7 @@ name = "程悦"
 def sigmoid(z):
     return 1.0 / (1 + np.exp(-z))
 
-def train_logreg(X, y, learning_rate=0.01, iterations=100):
+def train_logreg(X, y, learning_rate=0.01, iterations=1000, batch_size=32, regularization=None, lambda_reg=0.1):
     """
     对率回归训练函数。
 
@@ -40,12 +40,31 @@ def train_logreg(X, y, learning_rate=0.01, iterations=100):
     - numpy array: 训练得到的权重。
     """
     m, n = X.shape
-    weights = np.zeros(n)
+    weights = np.random.randn(n) * 0.01  # 随机初始化权重
     for i in range(iterations):
-        z = np.dot(X, weights)
+        indices = np.random.permutation(m)  # 随机打乱数据
-        h = sigmoid(z)
+        X_shuffled = X[indices]
-        gradient = np.dot(X.T, (h - y)) / m
+        y_shuffled = y[indices]
-        weights -= learning_rate * gradient
+
+        for j in range(0, m, batch_size):
+            X_batch = X_shuffled[j:j + batch_size]
+            y_batch = y_shuffled[j:j + batch_size]
+            z = np.dot(X_batch, weights)
+            h = sigmoid(z)
+            gradient = np.dot(X_batch.T, (h - y_batch)) / len(y_batch)
+
+            # 添加正则化项
+            if regularization == 'L1':
+                gradient += lambda_reg * np.sign(weights) / len(y_batch)
+            elif regularization == 'L2':
+                gradient += lambda_reg * weights / len(y_batch)
+
+            weights -= learning_rate * gradient
+
+        # 可选：学习率衰减
+        if i % 100 == 0:
+            learning_rate *= 0.95  # 每100次迭代衰减5%
+
     return weights
 
 def predict_logreg(inX, weights):
@@ -110,8 +129,8 @@ if __name__ == "__main__":
     X_test, y_test = load_data("./data/horse-colic.test")
 
     # 训练对率回归模型
-    weights = train_logreg(X_train, y_train, learning_rate=0.01, iterations=1000, re
+    weights = train_logreg(X_train, y_train, learning_rate=0.01, iterations=1000, batch_size=32, regularization='L2',
-    ='l1')
+                           lambda_reg=0.1)
 
     # 测试单个数据
     y_pred = predict_logreg(X_test[11], weights)