更新 function.py

function.py

@@ -1,79 +1,141 @@
-import numpy as np  
+import os
-  
+import numpy as np
-# 填写学号和姓名  
+
-id = "202210423105"  
+"""
-name = "魏明帅"  
+实验说明:
-  
+- 自行编写实现数据加载与预测病马死亡的对率回归算法。
-def sigmoid(z):  
+
-    return 1 / (1 + np.exp(-z))  
+注意事项:
-  
+- 在 function.py 中实现对率回归算法。
-def train_logreg(X, y, learning_rate=0.01, iterations=1000):  
+- 可以自行导入其他库，需要在 requirements.txt 添加依赖，若未在 requirements.txt 中添加使用到的依赖，则无法自动评测，成绩为零。
-    """  
+- 自动评测脚本仅会加载 train_logreg、predict_logreg 和 load_data 函数，请勿修改函数名。
-    对率回归训练函数。  
+
-  
+测评环境:
-    Args:  
+- CPU: Intel Xeon(R) CPU E3-1231 v3 @ 3.40GHz
-    - X (numpy array): 特征数据。  
+- OS: Ubuntu 22.04.4 LTS
-    - y (numpy array): 标签数据。  
+- Python: 3.10.12
-    - learning_rate (float): 学习率。  
+- numpy: 2.1.1
-    - iterations (int): 迭代次数。  
+"""
-  
+
-    Returns:  
+"""
-    - numpy array: 训练得到的权重。  
+请在这里填写学号和姓名，不填写即成绩无效。
-    """  
+"""
-    m, n = X.shape  
+id = "202210423105"
-    weights = np.zeros(n)  
+name = "魏明帅"
-    bias = 0  
+
-  
+# 这里写自己的代码实现
-    for i in range(iterations):  
+import numpy as np
-        h = sigmoid(np.dot(X, weights) + bias)  
+from sklearn.preprocessing import MinMaxScaler
-        error = h - y  
+from sklearn.model_selection import train_test_split
-        weights -= learning_rate * (1 / m) * np.dot(X.T, error)  
+
-        bias -= learning_rate * (1 / m) * np.sum(error)  
+
-  
+def sigmoid(z):
-    return weights, bias  
+    return 1.0 / (1 + np.exp(-z))
-  
+
-def predict_logreg(inX, weights, bias):  
+
-    """  
+def compute_gradient(X_batch, y_batch, weights, lambda_reg):
-    对率回归预测函数。  
+    z = np.dot(X_batch, weights)
-  
+    h = sigmoid(z)
-    Args:  
+    gradient = np.dot(X_batch.T, (h - y_batch)) / len(y_batch)
-    - inX (numpy array): 输入待分类的样本。  
+    gradient += (lambda_reg / len(y_batch)) * weights  # L2正则化
-    - weights (numpy array): 训练得到的权重。  
+    return gradient
-    - bias (float): 训练得到的偏置项。  
+
-  
+
-    Returns:  
+def train_logreg(X, y, learning_rate=0.001, iterations=10000, batch_size=64, lambda_reg=0.1):
-    - int: 预测的分类标签。  
+    m, n = X.shape
-    """  
+    weights = np.random.randn(n) * np.sqrt(2. / n)  # He初始化
-    prob = sigmoid(np.dot(inX, weights) + bias)  
+    best_weights = weights.copy()
-    return 1 if prob > 0.5 else 0  
+    best_accuracy = 0
-  
+
-def load_data(filepath):  
+    for _ in range(iterations):
-    """  
+        indices = np.random.permutation(m)
-    从文件中加载数据，若不自行正确实现，则无法通过评测，不记录成绩。  
+        X_shuffled = X[indices]
-  
+        y_shuffled = y[indices]
-    Args:  
+
-    - filepath (str): 数据文件路径。  
+        for j in range(0, m, batch_size):
-  
+            X_batch = X_shuffled[j:j + batch_size]
-    Returns:  
+            y_batch = y_shuffled[j:j + batch_size]
-    - np.array: 特征数据。  
+            gradient = compute_gradient(X_batch, y_batch, weights, lambda_reg)
-    - np.array: 标签数据。  
+            weights -= learning_rate * gradient
-    """  
+
-    data = np.loadtxt(filepath, delimiter=',')  
+        # 计算训练集准确率
-    X = data[:, :-1]  
+        train_accuracy = accuracy(X, y, weights)
-    y = data[:, -1]  
+        if train_accuracy > best_accuracy:
-    y = (y > 0).astype(int)  # 将标签转换为1和0  
+            best_accuracy = train_accuracy
-    return X, y  
+            best_weights = weights.copy()
-  
+
-if __name__ == "__main__":  
+        # 动态调整学习率
-    # 加载训练集  
+        if _ % 100 == 0 and train_accuracy > 0.95:  # 达到高准确率时减小学习率
-    X_train, y_train = load_data("./data/horse-colic.data")  
+            learning_rate *= 0.9
-  
+
-    # 加载测试集  
+    return best_weights
-    X_test, y_test = load_data("./data/horse-colic.test")  
+
-  
+
-    # 训练对率回归模型  
+def predict_logreg(X, weights):
-    weights, bias = train_logreg(X_train, y_train, learning_rate=0.01, iterations=1000)  
+    z = np.dot(X, weights)
-  
+    h = sigmoid(z)
-    # 测试单个数据  
+    return np.where(h >= 0.5, 1, 0)
-    y_pred = predict_logreg(X_test[11], weights, bias)  
+
-    print(f"预测结果: {y_pred}, 实际结果: {y_test[11]}")
+
+def load_data(filepath):
+    data, labels = [], []
+    with open(filepath, 'r') as f:
+        for line in f:
+            line = line.strip().split()
+            if len(line) >= 1 and line[-1].isdigit():
+                line_data = [0 if x == '?' else float(x) for x in line[:-1]]
+                data.append(line_data)
+                labels.append(int(line[-1]))
+
+    data = np.array(data)
+    labels = np.array(labels)
+
+    # 检查特征数量
+    num_features = data.shape[1]
+
+    # 特征选择：去掉特征3（索引2）和特征24~28（索引23~26），确保索引有效
+    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（索引23到26）
+
+    # 删除特征
+    data = np.delete(data, indices_to_delete, axis=1)
+
+    # 将标签二分类化，特征23（索引22）取值2和3合并为0，其余为1
+    target_variable = np.where((labels == 2) | (labels == 3), 0, 1)
+
+    return data, target_variable
+
+
+def accuracy(X, y, weights):
+    predictions = predict_logreg(X, weights)
+    return np.mean(predictions == y)
+
+
+# 其余代码保持不变
+if __name__ == "__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
+    val_accuracy = accuracy(X_val, y_val, final_weights)
+    print(f"验证集的准确率: {val_accuracy:.4f}")
+
+    # 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
+    final_test_accuracy = accuracy(X_test, y_test, final_weights)
+    print(f"最终测试集的准确率: {final_test_accuracy:.4f}")