更新 function.py

function.py

@@ -1,125 +1,102 @@
 import numpy as np
 import pandas as pd
 from sklearn.preprocessing import StandardScaler
-from sklearn.model_selection import train_test_split
+from sklearn.model_selection import train_test_split, KFold
+from sklearn.metrics import accuracy_score
 
 id = "202210423130"
 name = "董秋"
 
-
 def sigmoid(z):
-    """
-    Sigmoid 激活函数。
-    """
     return 1 / (1 + np.exp(-z))
 
 def train_logreg(X, y, learning_rate=0.01, iterations=1000, lambda_reg=0.01):
-    """
-    逻辑回归训练函数，随机初始化权重，并在每次迭代前打乱数据。
-    """
     m, n = X.shape
-    weights = np.random.randn(n) * 0.01  # 随机初始化权重，取较小值
-
+    weights = np.random.randn(n) * 0.01
     for i in range(iterations):
-        # 每次迭代打乱数据
         indices = np.random.permutation(m)
         X = X[indices]
         y = y[indices]
-
         z = np.dot(X, weights)
         predictions = sigmoid(z)
         gradient = np.dot(X.T, (predictions - y)) / m
-
-        # L2正则化
         gradient += (lambda_reg / m) * weights
-
-        # 更新权重
         weights -= learning_rate * gradient
-
-        # 动态调整学习率，逐渐减小
         learning_rate *= 0.99
-
     return weights
 
-
 def predict_logreg(inX, weights):
-    """
-    逻辑回归预测函数。
-    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)
     prediction_prob = sigmoid(z)
     return 1 if prediction_prob >= 0.5 else 0
 
 def handle_outliers(df, threshold=3):
-    """
-    使用 Z 分数法检测并处理异常值（噪音）。
-    任何 Z 分数大于指定阈值的数值会被替换为列的均值。
-    """
     z_scores = np.abs((df - df.mean()) / df.std())
-    df[z_scores > threshold] = np.nan  # 标记异常值为 NaN
-    df.fillna(df.mean(), inplace=True)  # 均值填充
+    df[z_scores > threshold] = np.nan
+    df.fillna(df.mean(), inplace=True)
     return df
 
-
 def load_data(filepath):
     data = []
     labels = []
-
     with open(filepath, 'r') as f:
         for line in f:
             values = line.strip().replace('?', '0').split()
             try:
                 values = [float(v) for v in values]
-                
-                # 合并特征23的取值
-                if values[22] == 2 or values[22] == 3:  # 特征23（索引22）
-                    labels.append(0)  # 死亡
+                if values[22] == 2 or values[22] == 3:
+                    labels.append(0)
                 elif values[22] == 1:
-                    labels.append(1)  # 生存
+                    labels.append(1)
                 else:
-                    continue  # 无效标签，丢弃
-                
-                # 用0填充缺失值
-                values = [np.nan if v == 0 else v for v in values]  # 先标记为NaN
-                data.append(values[:2] + values[3:23])  # 保留特征1, 2 和特征4-23
+                    continue
+                values = [np.nan if v == 0 else v for v in values]
+                data.append(values[:2] + values[3:23])
             except ValueError:
                 continue
-
     data = np.array(data)
     labels = np.array(labels)
-
-    # 转换成 DataFrame 以便进一步处理
     df = pd.DataFrame(data)
-
-    # 使用均值填充缺失值
-    df.fillna(df.mean(), inplace=True)  # 初步均值填充
-
-    # 处理噪音数据
-    df = handle_outliers(df)  # 检测并处理异常值
-
+    df.fillna(df.mean(), inplace=True)
+    df = handle_outliers(df)
     return df.to_numpy(), labels
 
 def preprocess_data(X):
-    # 特征标准化
     scaler = StandardScaler()
     return scaler.fit_transform(X)
 
+def cross_validate_learning_rate(X, y, learning_rates, k=5, iterations=1000, lambda_reg=0.01):
+    best_lr = None
+    best_score = 0
+    kf = KFold(n_splits=k, shuffle=True, random_state=42)
+
+    for lr in learning_rates:
+        cv_scores = []
+        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 = train_logreg(X_train, y_train, learning_rate=lr, iterations=iterations, lambda_reg=lambda_reg)
+            y_pred = [predict_logreg(x, weights) for x in X_val]
+            cv_scores.append(accuracy_score(y_val, y_pred))
+
+        avg_score = np.mean(cv_scores)
+        print(f"Learning Rate: {lr}, Cross-Validation Accuracy: {avg_score:.4f}")
+
+        if avg_score > best_score:
+            best_score = avg_score
+            best_lr = lr
+
+    print(f"Best Learning Rate: {best_lr}, Best CV Accuracy: {best_score:.4f}")
+    return best_lr
+
 if __name__ == "__main__":
-    # 加载数据集
     X, y = load_data("./data/horse-colic.data")
     X = preprocess_data(X)
 
-    # 使用 train_test_split 打乱并随机抽取数据
+    learning_rates = [0.001, 0.01, 0.05, 0.1, 0.2]
+    best_learning_rate = cross_validate_learning_rate(X, y, learning_rates)
+
     X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
-
-    # 训练逻辑回归模型
-    weights = train_logreg(X_train, y_train, learning_rate=0.1, iterations=1000, lambda_reg=0.01)
-
-    # 测试单个数据
+    weights = train_logreg(X_train, y_train, learning_rate=best_learning_rate, iterations=1000, lambda_reg=0.01)
     y_pred = predict_logreg(X_test[11], weights)
     print(f"预测结果: {y_pred}, 实际结果: {y_test[11]}")