更新 function.py

function.py

@@ -26,118 +26,92 @@ name = "廖运瑞"
 
 
 # 这里写自己的代码实现
-import numpy as np
+def train_logreg(X, y, learning_rate=0.01, iterations=1000):  
-from sklearn.preprocessing import MinMaxScaler
+    """  
-from sklearn.model_selection import train_test_split
+    对率回归训练函数。  
   
+    Args:  
+    - X (numpy array): 特征数据。  
+    - y (numpy array): 标签数据。  
+    - learning_rate (float): 学习率，默认为0.01。  
+    - iterations (int): 迭代次数，默认为1000。  
   
-def sigmoid(z):
+    Returns:  
-    return 1.0 / (1 + np.exp(-z))
+    - numpy array: 训练得到的权重。  
-
+    """  
-
-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 / len(y_batch)) * weights  # L2正则化
-    return gradient
-
-
-def train_logreg(X, y, learning_rate=0.001, iterations=10000, batch_size=64, lambda_reg=0.1):
     m, n = X.shape  
-    weights = np.random.randn(n) * np.sqrt(2. / n)  # He初始化
+    weights = np.zeros(n)  
-    best_weights = weights.copy()
+    bias = 0  # 偏置项，可以合并到权重中处理，但这里单独处理  
-    best_accuracy = 0
   
     for _ in range(iterations):  
-        indices = np.random.permutation(m)
+        output = sigmoid(np.dot(X, weights) + bias)  
-        X_shuffled = X[indices]
+        error = y - output  
-        y_shuffled = y[indices]
+        weights += learning_rate * np.dot(X.T, error) / m  
+        bias += learning_rate * np.sum(error) / m  
   
-        for j in range(0, m, batch_size):
+    # 如果需要返回偏置项，可以将其添加到权重数组的前面或作为单独的返回值  
-            X_batch = X_shuffled[j:j + batch_size]
+    # 这里为了与predict_logreg函数接口一致，我们选择不直接修改weights数组  
-            y_batch = y_shuffled[j:j + batch_size]
+    return weights, bias  
-            gradient = compute_gradient(X_batch, y_batch, weights, lambda_reg)
-            weights -= learning_rate * gradient
   
-        # 计算训练集准确率
+def predict_logreg(inX, weights, bias=0):  
-        train_accuracy = accuracy(X, y, weights)
+    """  
-        if train_accuracy > best_accuracy:
+    对率回归预测函数。  
-            best_accuracy = train_accuracy
-            best_weights = weights.copy()
   
-        # 动态调整学习率
+    Args:  
-        if _ % 100 == 0 and train_accuracy > 0.95:  # 达到高准确率时减小学习率
+    - inX (numpy array): 输入待分类的样本。  
-            learning_rate *= 0.9
+    - weights (numpy array): 训练得到的权重。  
+    - bias (float): 训练得到的偏置项，默认为0。  
   
-    return best_weights
+    Returns:  
+    - int: 预测的分类标签（0或1）。  
+    """  
+    z = np.dot(inX, weights) + bias  
+    return 1 if sigmoid(z) >= 0.5 else 0  
   
+def sigmoid(z):  
+    """  
+    Sigmoid激活函数。  
   
-def predict_logreg(X, weights):
+    Args:  
-    z = np.dot(X, weights)
+    - z (float or numpy array): 输入值。  
-    h = sigmoid(z)
-    return np.where(h >= 0.5, 1, 0)
   
+    Returns:  
+    - float or numpy array: Sigmoid输出值。  
+    """  
+    return 1 / (1 + np.exp(-z))  
   
 def load_data(filepath):  
-    data, labels = [], []
+    """  
-    with open(filepath, 'r') as f:
+    从文件中加载数据。  
-        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)
+    Args:  
-    labels = np.array(labels)
+    - filepath (str): 数据文件路径。  
   
-    # 检查特征数量
+    Returns:  
-    num_features = data.shape[1]
+    - np.array: 特征数据。  
+    - np.array: 标签数据。  
+    """  
+    data = np.loadtxt(filepath, delimiter=' ')  
+    X = data[:, 1:]  # 特征数据，假设从第二列开始  
+    y = data[:, 0]   # 标签数据，假设在第一列  
+    return X, y.astype(int)  # 确保标签为整数类型  
   
-    # 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
+    # 加载训练集  
-    X, y = load_data("./data/horse-colic.data")
+    X_train, y_train = load_data("./data/horse-colic.data")  
   
-    # 拆分训练集和验证集
+    # 为了处理偏置项，我们在X_train中添加一列全为1的列（或者使用上面的bias参数）  
-    X_train, X_val, y_train, y_val = train_test_split(X, y, test_size=0.2, random_state=42)
+    # 但为了与train_logreg函数接口一致，这里我们选择使用bias参数  
-
+    # 如果选择将偏置项合并到权重中，则应该这样处理：  
-    # 数据归一化
+    # X_train_with_bias = np.c_[np.ones((X_train.shape[0], 1)), X_train]  
-    scaler = MinMaxScaler()
+    # 并相应地修改train_logreg函数和predict_logreg函数的接口  
-    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
+    # 训练对率回归模型  
-    final_test_accuracy = accuracy(X_test, y_test, final_weights)
+    weights, bias = train_logreg(X_train, y_train, learning_rate=0.01, iterations=1000)  
-    print(f"最终测试集的准确率: {final_test_accuracy:.4f}")
+  
+    # 测试单个数据  
+    y_pred = predict_logreg(X_test[11], weights, bias)  
+    print(f"预测结果: {y_pred}, 实际结果: {y_test[11]}")