更新 function.py

function.py

@@ -1,6 +1,7 @@
+# function.py
 import numpy as np
 from sklearn.preprocessing import StandardScaler
-from sklearn.model_selection import KFold
+from itertools import product
 
 """
 实验说明:
@@ -24,78 +25,86 @@ from sklearn.model_selection import KFold
 id = "202210423117"
 name = "程悦"
 
+
 def sigmoid(z):
-    """计算Sigmoid激活函数值"""
     return 1.0 / (1 + np.exp(-z))
 
 
-def train_logreg(X, y, learning_rate=0.05, iterations=1500, batch_size=64,
-                 regularization='L2', lambda_reg=0.05, cv=5):
-    """训练逻辑回归模型并使用交叉验证计算权重"""
+def train_logreg(X, y, learning_rate=0.05, iterations=1500, batch_size=64, regularization='L2', lambda_reg=0.05):
+    """
+    对率回归训练函数。
+
+    Args:
+    - X (numpy array): 特征数据。
+    - y (numpy array): 标签数据。
+
+    Returns:
+    - numpy array: 训练得到的权重。
+    """
     m, n = X.shape
-    kf = KFold(n_splits=cv, shuffle=True)
-    weights_list = []
-    fold_scores = []
+    weights = np.random.randn(n) * 0.01  # 随机初始化权重
+    for i in range(iterations):
+        indices = np.random.permutation(m)  # 随机打乱数据
+        X_shuffled = X[indices]
+        y_shuffled = y[indices]
 
-    for train_index, val_index in kf.split(X):
-        X_train, X_val = X[train_index], X[val_index]
-        y_train, y_val = y[train_index], y[val_index]
-        weights = np.random.randn(n) * 0.01
+        for j in range(0, m, batch_size):
+            X_batch = X_shuffled[j:j + batch_size]
+            y_batch = y_shuffled[j:j + batch_size]
+            z = np.dot(X_batch, weights)
+            h = sigmoid(z)
+            gradient = np.dot(X_batch.T, (h - y_batch)) / len(y_batch)
 
-        for _ in range(iterations):
-            indices = np.random.permutation(len(X_train))
-            X_shuffled, y_shuffled = X_train[indices], y_train[indices]
+            # 添加正则化项
+            if regularization == 'L1':
+                gradient += lambda_reg * np.sign(weights) / len(y_batch)
+            elif regularization == 'L2':
+                gradient += lambda_reg * weights / len(y_batch)
 
-            for j in range(0, len(X_train), batch_size):
-                X_batch = X_shuffled[j:j + batch_size]
-                y_batch = y_shuffled[j:j + batch_size]
-                z = np.dot(X_batch, weights)
-                h = sigmoid(z)
-                gradient = np.dot(X_batch.T, (h - y_batch)) / len(y_batch)
+            weights -= learning_rate * gradient
 
-                # 正则化处理
-                if regularization == 'L1':
-                    gradient += lambda_reg * np.sign(weights) / len(y_batch)
-                elif regularization == 'L2':
-                    gradient += (lambda_reg / len(y_train)) * weights
+        # 可选：学习率衰减
+        if i % 100 == 0:
+            learning_rate *= 0.95  # 每100次迭代衰减5%
 
-                weights -= learning_rate * gradient
-
-            if _ % 100 == 0:
-                learning_rate *= 0.95  # 动态调整学习率
-
-        # 验证集准确率
-        predictions = predict_logreg(X_val, weights)
-        score = np.mean(predictions == y_val)
-        fold_scores.append(score)
-        weights_list.append(weights)
-
-    # 权重加权平均
-    total_score = sum(fold_scores)
-    if total_score == 0:
-        return weights_list[-1]
-
-    averaged_weights = np.zeros(n)
-    for weight, score in zip(weights_list, fold_scores):
-        averaged_weights += weight * score
-    return averaged_weights / total_score
+    return weights
 
 
-def predict_logreg(X, weights):
-    """对输入数据进行逻辑回归预测"""
-    z = np.dot(X, weights)
+def predict_logreg(inX, weights):
+    """
+    对率回归预测函数。
+
+    Args:
+    - inX (numpy array): 输入待分类的样本。
+    - weights (numpy array): 训练得到的权重。
+
+    Returns:
+    - int: 预测的分类标签。
+    """
+    z = np.dot(inX, weights)
     h = sigmoid(z)
-    return (h > 0.5).astype(int)
+    return 1 if h > 0.5 else 0
 
 
 def load_data(filepath):
-    """加载数据并处理缺失值"""
-    data, labels = [], []
+    """
+    从文件中加载数据，填充缺失值，并进行特征处理。
 
+    Args:
+    - filepath (str): 数据文件路径。
+
+    Returns:
+    - np.array: 特征数据。
+    - np.array: 标签数据。
+    """
+    data = []
+    labels = []
     with open(filepath, 'r') as f:
-        for line in f:
+        lines = f.readlines()
+        for line in lines:
             line = line.strip().split()
-            if line and line[-1].isdigit():
+            if len(line) >= 1 and line[-1].isdigit():
+                # 将缺失值 '?' 填充为 0，并转换为浮点数
                 line_data = [1] + [0 if x == '?' else float(x) for x in line[:-1]]
                 data.append(line_data)
                 labels.append(int(line[-1]))
@@ -103,47 +112,25 @@ def load_data(filepath):
     data = np.array(data)
     labels = np.array(labels)
 
-    # 处理目标变量
-    labels[(labels == 2) | (labels == 3)] = 0
-    data = np.delete(data, [2] + list(range(23, 28)), axis=1)
+    # 选择特征，合并目标变量
+    target_variable = labels
+    target_variable[(target_variable == 2) | (target_variable == 3)] = 0  # 合并死亡和安乐死为0
+
+    # 去掉特征3和特征24~28
+    data = np.delete(data, [2] + list(range(23, 28)), axis=1)  # 特征3的索引是2，特征24-28的索引是23-27
 
-    # 数据标准化
     scaler = StandardScaler()
     data = scaler.fit_transform(data)
 
-    return data, labels
-
-
-def accuracy(X, y, weights):
-    """计算模型的准确率"""
-    predictions = predict_logreg(X, weights)
-    return np.mean(predictions == y)
-
-
-def ensemble_predict(X, models, weights):
-    """集成模型预测"""
-    predictions = np.array([predict_logreg(X, model) for model in models])  # (num_models, num_samples)
-    weighted_predictions = np.tensordot(predictions, weights, axes=(0, 0))  # 权重加权
-    return (weighted_predictions > 0.5).astype(int)  # 返回最终预测
-
+    return data, target_variable
 
 
 if __name__ == "__main__":
     X_train, y_train = load_data("./data/horse-colic.data")
     X_test, y_test = load_data("./data/horse-colic.test")
 
-    # 训练多个模型
-    num_models = 5
-    models = [train_logreg(X_train, y_train, learning_rate=0.02, iterations=6000,
-                           batch_size=48, regularization='None', lambda_reg=0.02)
-              for _ in range(num_models)]
+    final_weights = train_logreg(X_train, y_train, learning_rate=0.05, iterations=1500, batch_size=64, regularization='L2', lambda_reg=0.05)
 
-    # 模型准确率作为权重
-    model_accuracies = [accuracy(X_train, y_train, model) for model in models]
-    weights = np.array(model_accuracies) / sum(model_accuracies)
-
-    # 集成预测
-    ensemble_predictions = ensemble_predict(X_test, models, weights)
-
-    final_accuracy = np.mean(ensemble_predictions == y_test)
-    print(f"最终测试集的准确率: {final_accuracy:.4f}")
+    # 计算并输出最终测试集的准确率
+    final_accuracy = accuracy(X_test, y_test, final_weights)
+    print(f"最终测试集的准确率: {final_accuracy:.4f}")