diff --git a/function.py b/function.py
index 02b691a..ba8d467 100644
--- a/function.py
+++ b/function.py
@@ -35,20 +35,26 @@ def load_data(filepath):
     y = data[:, -1]
     return X, y
 
-def train_logreg(X, y, learning_rate=0.01, iterations=1000):
-    """训练逻辑回归模型，使用随机梯度下降."""
+def train_logreg(X, y, initial_learning_rate=0.1, iterations=1000, decay_rate=0.01):
+    """训练逻辑回归模型，使用随机梯度下降和学习率衰减."""
     n_samples, n_features = X.shape
     weights = np.zeros(n_features + 1)  # 增加一位用于偏置
     X = np.c_[X, np.ones((n_samples, 1))]  # 增加偏置项
 
-    for _ in range(iterations):
+    for i in range(iterations):
+        # 更新学习率
+        learning_rate = initial_learning_rate / (1 + decay_rate * i)
+
         random_index = np.random.randint(n_samples)
         X_i = X[random_index]
         y_i = y[random_index]
         
         linear_model = np.dot(X_i, weights)
-        dw = (linear_model - y_i) * X_i
-        weights -= learning_rate * dw
+        prediction = 1 / (1 + np.exp(-linear_model))  # Sigmoid函数
+        error = prediction - y_i
+        
+        # 更新权重
+        weights -= learning_rate * error * X_i
 
     return weights
 
@@ -56,7 +62,8 @@ def predict_logreg(inX, weights):
     """使用线性模型预测结果."""
     inX = np.append(inX, 1)  # 增加偏置项
     linear_model = np.dot(inX, weights)
-    return 1.0 if linear_model >= 0 else 2.0 
+    prediction = 1 / (1 + np.exp(-linear_model))
+    return 1.0 if prediction >= 0.5 else 2.0  # 返回1.0或2.0
 
 if __name__ == "__main__":
     # 加载训练集