更新 function.py

function.py

@@ -1,67 +1,63 @@
 import numpy as np
 
-# 学号和姓名  
 id = "202210423224"
 name = "吴杨洋"
 
 def sigmoid(z):
-    """  
-    Sigmoid 激活函数。  
-    """  
+    """Sigmoid 激活函数，添加 clipping 防止溢出"""
+    z = np.clip(z, -500, 500)  # 限制 z 的值范围
     return 1 / (1 + np.exp(-z))
 
-def train_logreg(X, y, num_iter=1000, learning_rate=0.01, lambda_reg=0.01):  
+def train_logreg(X, y, numIter=1000, learning_rate=0.01, lambda_reg=0.01, batch_size=64):
     m, n = X.shape
     weights = np.zeros(n)
-
-    for _ in range(num_iter):  
-        z = np.dot(X, weights)  
+    for j in range(numIter):
+        indices = np.random.permutation(m)
+        X_shuffled = X[indices]
+        y_shuffled = y[indices]
+        for i in range(0, m, batch_size):
+            end = min(i + batch_size, m)
+            batch_X = X_shuffled[i:end]
+            batch_y = y_shuffled[i:end]
+            z = np.dot(batch_X, weights)
             h = sigmoid(z)
-        error = y - h  
-        
-        # 计算梯度并添加正则化项  
-        gradient = np.dot(X.T, error) / m + (lambda_reg / m) * weights  
-        weights += learning_rate * gradient  
-
+            error = batch_y - h
+            gradient = np.dot(batch_X.T, error) / len(batch_y)
+            reg_term = lambda_reg * weights
+            gradient += reg_term / len(batch_y)
+            weights -= learning_rate * gradient
     return weights
 
-def predict_logreg(X, weights):  
-    """  
-    对率回归预测函数。  
-    
-    Args:  
-    - X (numpy array): 输入待分类的样本。  
-    - weights (numpy array): 训练得到的权重。  
-    
-    Returns:  
-    - numpy array: 预测的分类标签。  
-    """  
-    probabilities = sigmoid(np.dot(X, weights))  
-    return (probabilities >= 0.5).astype(int)  
+def predict_logreg(inX, weights):
+    z = np.dot(inX, weights)
+    prob = sigmoid(z)
+    return 1 if prob >= 0.5 else 0
 
 def load_data(filepath):
-    """  
-    从文件中加载数据。  
-
-    Args:  
-    - filepath (str): 数据文件路径。  
-
-    Returns:  
-    - X (np.array): 特征数据。  
-    - y (np.array): 标签数据。  
-    """  
-    data = np.genfromtxt(filepath, delimiter=',', dtype=str, filling_values=np.nan)  
-    
-    # 处理缺失值和类型转换  
-    data = np.where(data == '?', np.nan, data).astype(float)  
-    data = data[~np.isnan(data).any(axis=1)]  # 删除含有NaN的行  
-
-    # 提取特征和标签  
-    X = np.delete(data, [22, 24, 25, 26, -1], axis=1)  # 删除不需要的列  
-    y = np.array([1 if row[22] in ['2', '3'] else 0 for row in data])  # 生成标签  
-    X = np.insert(X, 0, 1, axis=1)  # 添加偏置项  
-
-    return X, y  
+    file = open(filepath)
+    fr = file.readlines()
+    data_list = []
+    y = []
+    for data in fr:
+        data = data.strip()
+        data = data.split(' ')
+        data_25 = data[24]
+        y.append(data[22])
+        del data[-1]
+        del data[26]
+        del data[25]
+        del data[24]
+        del data[22]
+        del data[2]
+        for i in data_25:
+            data.append(i)
+        data = [0 if data_elem == '?' else float(data_elem) for data_elem in data]
+        data_list.append(data)
+    data_number = len(y)
+    y_train = np.array([1 if y[j] == '2' or y[j] == '3' else 0 for j in range(data_number)])
+    X = np.array(data_list)
+    X_train = np.insert(X, 0, 1, axis=1)  # 添加偏置项
+    return X_train, y_train
 
 def standardize(X):
     mean = X.mean(axis=0)
@@ -69,19 +65,15 @@ def standardize(X):
     return (X - mean) / std, mean, std
 
 def main():
-    # 加载训练集和测试集  
     X_train, y_train = load_data("./data/horse-colic.data")
     X_test, y_test = load_data("./data/horse-colic.test")
 
-    # 数据标准化  
     X_train, mean, std = standardize(X_train)
     X_test = (X_test - mean) / std
 
-    # 训练对率回归模型  
-    weights = train_logreg(X_train, y_train, num_iter=1000, learning_rate=0.01, lambda_reg=0.1)  
+    weights = train_logreg(X_train, y_train, numIter=1000, learning_rate=0.01, lambda_reg=0.1, batch_size=64)
 
-    # 测试模型  
-    y_pred_test = predict_logreg(X_test, weights)  
+    y_pred_test = [predict_logreg(x, weights) for x in X_test]
     accuracy_test = np.mean(y_pred_test == y_test)
     print(f"Test Accuracy: {accuracy_test:.2f}")