更新 function.py

function.py

@@ -31,77 +31,88 @@ 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
+# Sigmoid函数，用于逻辑回归模型
 def sigmoid(z):
     return 1.0 / (1 + np.exp(-z))
 
-# compute_gradient函数，用于计算逻辑回归的梯度
+# 计算梯度，用于逻辑回归的梯度下降算法
 def compute_gradient(X_batch, y_batch, weights, lambda_reg):
-    z = np.dot(X_batch, weights)  # 计算当前权重下的线性组合
+    z = np.dot(X_batch, weights)  # 计算z值
-    h = sigmoid(z)  # 应用sigmoid函数将线性组合转换为概率
+    h = sigmoid(z)  # 计算sigmoid函数值
     gradient = np.dot(X_batch.T, (h - y_batch)) / len(y_batch)  # 计算梯度
     gradient += (lambda_reg / len(y_batch)) * weights  # 添加L2正则化项
     return gradient
 
-# train_logreg函数，用于训练逻辑回归模型
+# 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):
     m, n = X.shape  # 获取样本数量和特征数量
     weights = np.random.randn(n) * np.sqrt(2. / n)  # 使用He初始化权重
     best_weights = weights.copy()  # 存储最佳权重
     best_accuracy = 0  # 存储最佳准确率
 
-    for i in range(iterations):  # 迭代指定次数
+    for _ in range(iterations):  # 迭代指定次数
         indices = np.random.permutation(m)  # 随机打乱数据
-        X_shuffled = X[indices]
+        X_shuffled = X[indices]  # 打乱特征
-        y_shuffled = y[indices]
+        y_shuffled = y[indices]  # 打乱标签
 
-        for j in range(0, m, batch_size):  # 批量梯度下降
+        for j in range(0, m, batch_size):  # 按批次处理数据
-            X_batch = X_shuffled[j:j + batch_size]
+            X_batch = X_shuffled[j:j + batch_size]  # 获取当前批次的特征
-            y_batch = y_shuffled[j:j + batch_size]
+            y_batch = y_shuffled[j:j + batch_size]  # 获取当前批次的标签
             gradient = compute_gradient(X_batch, y_batch, weights, lambda_reg)  # 计算梯度
             weights -= learning_rate * gradient  # 更新权重
 
-        accuracy = accuracy_score(y, predict_logreg(X, weights))  # 计算当前准确率
+        # 计算训练集准确率
-        if accuracy > best_accuracy:  # 更新最佳准确率和最佳权重
+        train_accuracy = accuracy(X, y, weights)
-            best_accuracy = accuracy
+        if train_accuracy > best_accuracy:  # 如果当前准确率更高，则更新最佳权重和准确率
+            best_accuracy = train_accuracy
             best_weights = weights.copy()
 
-        if i % 1000 == 0 and accuracy > 0.95:  # 动态调整学习率
+        # 动态调整学习率
+        if _ % 100 == 0 and train_accuracy > 0.95:  # 每100次迭代检查一次，如果准确率超过0.95，则减小学习率
             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
+# 预测函数，用于逻辑回归模型
 def predict_logreg(X, weights):
-    z = np.dot(X, weights)  # 计算线性组合
+    z = np.dot(X, weights)  # 计算z值
-    h = sigmoid(z)  # 应用sigmoid函数
+    h = sigmoid(z)  # 计算sigmoid函数值
     return np.where(h >= 0.5, 1, 0)  # 返回预测结果
 
-# preprocess_data函数，用于预处理数据
+# 加载数据函数
-def preprocess_data(X, y):
-    y[y == 2] = 0  # 将类别2转换为0
-    y[y == 3] = 0  # 将类别3转换为0
-    X = np.delete(X, [3], axis=1)  # 删除特征3
-    X = X[:, :23]  # 删除特征24~28
-    return X, y
-
-# load_data函数，用于加载数据
 def load_data(filepath):
-    try:
+    data, labels = [], []
-        data = pd.read_csv(filepath, header=None, sep=r'\s+', on_bad_lines='skip')  # 读取数据
+    with open(filepath, 'r') as f:  # 打开文件
-    except pd.errors.ParserError as e:
+        for line in f:  # 逐行读取
-        print(f"Error parsing file: {e}")
+            line = line.strip().split()  # 去除空格并分割
-        return None, None
+            if len(line) >= 1 and line[-1].isdigit():  # 确保行数据有效
-    data.replace('?', np.nan, inplace=True)  # 将'?'替换为np.nan
+                line_data = [0 if x == '?' else float(x) for x in line[:-1]]  # 将非数字替换为0并转换为浮点数
-    imputer = SimpleImputer(missing_values=np.nan, strategy='constant', fill_value=0)  # 填充缺失值
+                data.append(line_data)
-    data.iloc[:, :-1] = imputer.fit_transform(data.iloc[:, :-1])
+                labels.append(int(line[-1]))  # 将标签转换为整数
-    for col in data.columns[:-1]:
-        data[col] = pd.to_numeric(data[col], errors='coerce')  # 转换为数值类型
-    X = data.iloc[:, :-1].values  # 特征数据
-    y = data.iloc[:, -1].values.astype(int)  # 目标变量
-    X, y = preprocess_data(X, y)  # 预处理数据
-    return X, y
 
+    data = np.array(data)  # 转换为numpy数组
+    labels = np.array(labels)
+
+    # 检查特征数量并进行特征选择
+    num_features = data.shape[1]
+    indices_to_delete = [2]  # 特征3
+    if num_features > 24:  # 确保特征数量足够
+        indices_to_delete += list(range(24, 27))  # 特征24到27
+
+    # 删除特征
+    data = np.delete(data, indices_to_delete, axis=1)
+
+    # 将标签二分类化
+    target_variable = np.where((labels == 2) | (labels == 3), 0, 1)
+
+    return data, target_variable
+
+# 计算准确率函数
+def accuracy(X, y, weights):
+    predictions = predict_logreg(X, weights)  # 预测结果
+    return np.mean(predictions == y)  # 计算准确率
+
+# 主函数
 if __name__ == "__main__":
     # 加载数据
     X, y = load_data("./data/horse-colic.data")
@@ -118,8 +129,7 @@ if __name__ == "__main__":
     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(X_val, y_val, final_weights)
-    val_accuracy = accuracy_score(y_val, y_pred)
     print(f"验证集的准确率: {val_accuracy:.4f}")
 
     # 加载测试集
@@ -127,6 +137,5 @@ if __name__ == "__main__":
     X_test = scaler.transform(X_test)
 
     # 测试模型
-    y_test_pred = predict_logreg(X_test, final_weights)
+    final_test_accuracy = accuracy(X_test, y_test, final_weights)
-    test_accuracy = accuracy_score(y_test, y_test_pred)
+    print(f"最终测试集的准确率: {final_test_accuracy:.4f}")
-    print(f"最终测试集的准确率: {test_accuracy:.4f}")