更新 function.py

function.py

@@ -31,24 +31,21 @@ name = "严锡好"
 
 
 # 这里写自己的代码实现
-# sigmoid函数，用于逻辑回归模型中将线性输出转换为概率
+def sigmoid(z):
-def sigmoid(x):
+    return 1.0 / (1 + np.exp(-z))
-    return 1 / (1 + np.exp(-x))
 
-# 计算梯度的函数
 def compute_gradient(X_batch, y_batch, weights, lambda_reg):
     z = np.dot(X_batch, weights)
     h = sigmoid(z)
     gradient = np.dot(X_batch.T, (h - y_batch)) / len(y_batch)
-    gradient += lambda_reg * weights  # L2正则化
+    gradient += (lambda_reg / len(y_batch)) * weights  # L2正则化
     return gradient
 
-# train_logreg函数，用于训练逻辑回归模型
+def train_logreg(X, y, learning_rate=0.001, iterations=10000, batch_size=64, lambda_reg=0.1):
-def train_logreg(X, y, learning_rate=0.01, iterations=1000, batch_size=64, lambda_reg=0.1):
     m, n = X.shape
-    weights = np.zeros(n)
+    weights = np.random.randn(n) * np.sqrt(2. / n)  # He初始化
-    best_accuracy = 0
     best_weights = weights.copy()
+    best_accuracy = 0
 
     for i in range(iterations):
         indices = np.random.permutation(m)
@@ -61,25 +58,21 @@ def train_logreg(X, y, learning_rate=0.01, iterations=1000, batch_size=64, lambd
             gradient = compute_gradient(X_batch, y_batch, weights, lambda_reg)
             weights -= learning_rate * gradient
 
-        # 计算训练集准确率
         accuracy = accuracy_score(y, predict_logreg(X, weights))
         if accuracy > best_accuracy:
             best_accuracy = accuracy
             best_weights = weights.copy()
 
-        # 动态调整学习率
+        if i % 1000 == 0 and accuracy > 0.95:  # 动态调整学习率
-        if i % 100 == 0 and accuracy > 0.95:  # 达到高准确率时减小学习率
             learning_rate *= 0.9
 
     return best_weights
 
-# predict_logreg函数，用于使用训练好的权重进行预测
+def predict_logreg(X, weights):
-def predict_logreg(inX, weights):
+    z = np.dot(X, weights)
-    linear_output = np.dot(inX, weights)
+    h = sigmoid(z)
-    probability = sigmoid(linear_output)
+    return np.where(h >= 0.5, 1, 0)
-    return np.where(probability > 0.5, 1, 0)
 
-# preprocess_data函数，用于预处理数据
 def preprocess_data(X, y):
     y[y == 2] = 0
     y[y == 3] = 0
@@ -87,7 +80,6 @@ def preprocess_data(X, y):
     X = X[:, :23]
     return X, y
 
-# load_data函数，用于加载数据
 def load_data(filepath):
     try:
         data = pd.read_csv(filepath, header=None, sep=r'\s+', on_bad_lines='skip')
@@ -104,28 +96,33 @@ def load_data(filepath):
     X, y = preprocess_data(X, y)
     return X, y
 
-# main函数，程序的入口点
-def main():
-    X, y = load_data("./data/horse-colic.data")
-    if X is not None:
-        # 分割数据集为训练集和测试集
-        X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
-        
-        # 特征缩放
-        scaler = StandardScaler()
-        X_train = scaler.fit_transform(X_train)
-        X_test = scaler.transform(X_test)
-        
-        # 训练模型
-        weights = train_logreg(X_train, y_train, learning_rate=0.01, iterations=1000, batch_size=64, lambda_reg=0.1)
-        
-        # 预测测试集
-        y_pred = predict_logreg(X_test, weights)
-        
-        # 计算准确率
-        accuracy = accuracy_score(y_test, y_pred)
-        print(f"准确率: {accuracy}")
-        print(classification_report(y_test, y_pred))
-
 if __name__ == "__main__":
-    main()
+    # 加载数据
+    X, y = load_data("./data/horse-colic.data")
+
+    # 拆分训练集和验证集
+    X_train, X_val, y_train, y_val = train_test_split(X, y, test_size=0.2, random_state=42)
+
+    # 数据归一化
+    scaler = MinMaxScaler()
+    X_train = scaler.fit_transform(X_train)
+    X_val = scaler.transform(X_val)
+
+    # 训练模型
+    final_weights = train_logreg(X_train, y_train, learning_rate=0.01, iterations=5000, batch_size=64, lambda_reg=0.1)
+
+    # 验证模型
+    y_pred = predict_logreg(X_val, final_weights)
+    val_accuracy = accuracy_score(y_val, y_pred)
+    print(f"验证集的准确率: {val_accuracy:.4f}")
+    print(classification_report(y_val, y_pred))
+
+    # 加载测试集
+    X_test, y_test = load_data("./data/horse-colic.test")
+    X_test = scaler.transform(X_test)
+
+    # 测试模型
+    y_test_pred = predict_logreg(X_test, final_weights)
+    test_accuracy = accuracy_score(y_test, y_test_pred)
+    print(f"最终测试集的准确率: {test_accuracy:.4f}")
+    print(classification_report(y_test, y_test_pred))