更新 function.py

function.py

@@ -1,9 +1,10 @@
 # function.py
 import numpy as np
 import pandas as pd
+from sklearn.model_selection import train_test_split, KFold
+from sklearn.preprocessing import MinMaxScaler
 from sklearn.impute import SimpleImputer
 from sklearn.preprocessing import StandardScaler
-from sklearn.model_selection import train_test_split
 from sklearn.metrics import accuracy_score, classification_report
 
 """
@@ -31,88 +32,105 @@ name = "严锡好"
 
 
 # 这里写自己的代码实现
+# Sigmoid函数，用于逻辑回归模型
 def sigmoid(z):
     return 1.0 / (1 + np.exp(-z))
 
+# 计算梯度，用于逻辑回归的梯度下降算法
 def compute_gradient(X_batch, y_batch, weights, lambda_reg):
-    z = np.dot(X_batch, weights)
+    z = np.dot(X_batch, weights)  # 计算z值
-    h = sigmoid(z)
+    h = sigmoid(z)  # 计算sigmoid函数值
-    gradient = np.dot(X_batch.T, (h - y_batch)) / len(y_batch)
+    gradient = np.dot(X_batch.T, (h - y_batch)) / len(y_batch)  # 计算梯度
-    gradient += (lambda_reg / len(y_batch)) * weights
+    gradient += (lambda_reg / len(y_batch)) * weights  # 添加L2正则化项
     return gradient
 
 # Adam优化器更新函数
 def adam_update(grad, m, v, t, lr=0.001, beta1=0.9, beta2=0.999, epsilon=1e-8):
-    m = beta1 * m + (1 - beta1) * grad
+    m = beta1 * m + (1 - beta1) * grad  # 更新一阶矩估计
-    v = beta2 * v + (1 - beta2) * (grad ** 2)
+    v = beta2 * v + (1 - beta2) * (grad ** 2)  # 更新二阶矩估计
-    m_hat = m / (1 - beta1 ** t)
+    m_hat = m / (1 - beta1 ** t)  # 计算偏差校正的一阶矩
-    v_hat = v / (1 - beta2 ** t)
+    v_hat = v / (1 - beta2 ** t)  # 计算偏差校正的二阶矩
-    return lr * m_hat / (np.sqrt(v_hat) + epsilon), m, v
+    return lr * m_hat / (np.sqrt(v_hat) + epsilon), m, v  # 返回更新步长和更新后的一阶矩和二阶矩
 
+# 训练逻辑回归模型，使用Adam优化器
 def train_logreg(X, y, learning_rate=0.001, iterations=10000, batch_size=64, lambda_reg=0.1):
-    m, n = X.shape
+    m, n = X.shape  # 获取样本数量和特征数量
-    weights = np.random.randn(n) * np.sqrt(2. / n)
+    weights = np.random.randn(n) * np.sqrt(2. / n)  # 使用He初始化权重
-    m_t, v_t = np.zeros(n), np.zeros(n)
+    m_t, v_t = np.zeros(n), np.zeros(n)  # 初始化Adam优化器的一阶矩和二阶矩
-    best_weights = weights.copy()
+    best_weights = weights.copy()  # 存储最佳权重
-    best_accuracy = 0
+    best_accuracy = 0  # 存储最佳准确率
 
-    for epoch in range(iterations):
+    for epoch in range(iterations):  # 迭代指定次数
-        indices = np.random.permutation(m)
+        indices = np.random.permutation(m)  # 随机打乱数据
-        X_shuffled, y_shuffled = X[indices], y[indices]
+        X_shuffled, y_shuffled = X[indices], y[indices]  # 打乱特征和标签
 
-        for j in range(0, m, batch_size):
+        for j in range(0, m, batch_size):  # 按批次处理数据
-            X_batch, y_batch = X_shuffled[j:j + batch_size], y_shuffled[j:j + batch_size]
+            X_batch, y_batch = X_shuffled[j:j + batch_size], y_shuffled[j:j + batch_size]  # 获取当前批次的特征和标签
-            grad = compute_gradient(X_batch, y_batch, weights, lambda_reg)
+            grad = compute_gradient(X_batch, y_batch, weights, lambda_reg)  # 计算梯度
-            update, m_t, v_t = adam_update(grad, m_t, v_t, epoch + 1, learning_rate)
+            update, m_t, v_t = adam_update(grad, m_t, v_t, epoch + 1, learning_rate)  # 使用Adam优化器更新权重
-            weights -= update
+            weights -= update  # 更新权重
 
-        train_accuracy = accuracy(X, y, weights)
+        train_accuracy = accuracy(X, y, weights)  # 计算训练集准确率
-        if train_accuracy > best_accuracy:
+        if train_accuracy > best_accuracy:  # 如果当前准确率更高，则更新最佳权重和准确率
             best_accuracy = train_accuracy
             best_weights = weights.copy()
 
-        if epoch % 100 == 0 and train_accuracy > 0.95:
+        if epoch % 100 == 0 and train_accuracy > 0.95:  # 每100次迭代检查一次，如果准确率超过0.95，则减小学习率
             learning_rate *= 0.9
 
     return best_weights
 
+# 预测函数，用于逻辑回归模型
 def predict_logreg(X, weights):
-    z = np.dot(X, weights)
+    z = np.dot(X, weights)  # 计算z值
-    h = sigmoid(z)
+    h = sigmoid(z)  # 计算sigmoid函数值
-    return np.where(h >= 0.5, 1, 0)
+    return np.where(h >= 0.5, 1, 0)  # 返回预测结果
 
+# 加载数据函数
 def load_data(filepath):
     data, labels = [], []
-    with open(filepath, 'r') as f:
+    with open(filepath, 'r') as f:  # 打开文件
-        for line in f:
+        for line in f:  # 逐行读取
-            line = line.strip().split()
+            line = line.strip().split()  # 去除空格并分割
-            if len(line) >= 1 and line[-1].isdigit():
+            if len(line) >= 1 and line[-1].isdigit():  # 确保行数据有效
-                line_data = [0 if x == '?' else float(x) for x in line[:-1]]
+                line_data = [0 if x == '?' else float(x) for x in line[:-1]]  # 将非数字替换为0并转换为浮点数
                 data.append(line_data)
-                labels.append(int(line[-1]))
+                labels.append(int(line[-1]))  # 将标签转换为整数
 
-    data, labels = np.array(data), np.array(labels)
+    data, labels = np.array(data), np.array(labels)  # 转换为numpy数组
-    indices_to_delete = [2] + (list(range(24, 27)) if data.shape[1] > 24 else [])
+
+    # 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 data.shape[1] > 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
 
+# 计算准确率函数
 def accuracy(X, y, weights):
-    predictions = predict_logreg(X, weights)
+    predictions = predict_logreg(X, weights)  # 预测结果
-    return np.mean(predictions == y)
+    return np.mean(predictions == y)  # 计算准确率
 
+# 主函数
 if __name__ == "__main__":
-    X, y = load_data("./data/horse-colic.data")
+    X, y = load_data("./data/horse-colic.data")  # 加载数据
-    X_train, X_val, y_train, y_val = train_test_split(X, y, test_size=0.2, random_state=42)
+    X_train, X_val, y_train, y_val = train_test_split(X, y, test_size=0.2, random_state=42)  # 拆分训练集和验证集
 
-    scaler = MinMaxScaler()
+    scaler = MinMaxScaler()  # 初始化归一化对象
-    X_train, X_val = scaler.fit_transform(X_train), scaler.transform(X_val)
+    X_train = scaler.fit_transform(X_train)  # 训练集归一化
+    X_val = scaler.transform(X_val)  # 验证集归一化
 
-    final_weights = train_logreg(X_train, y_train, learning_rate=0.01, iterations=5000, batch_size=64, lambda_reg=0.1)
+    final_weights = train_logreg(X_train, y_train, learning_rate=0.01, iterations=5000, batch_size=64, lambda_reg=0.1)  # 训练模型
-    val_accuracy = accuracy(X_val, y_val, final_weights)
+    val_accuracy = accuracy(X_val, y_val, final_weights)  # 验证模型
-    print(f"验证集的准确率: {val_accuracy:.4f}")
+    print(f"验证集的准确率: {val_accuracy:.4f}")  # 打印验证集准确率
 
-    X_test, y_test = load_data("./data/horse-colic.test")
+    X_test, y_test = load_data("./data/horse-colic.test")  # 加载测试集
-    X_test = scaler.transform(X_test)
+    X_test = scaler.transform(X_test)  # 测试集归一化
-    final_test_accuracy = accuracy(X_test, y_test, final_weights)
+    final_test_accuracy = accuracy(X_test, y_test, final_weights)  # 测试模型
-    print(f"最终测试集的准确率: {final_test_accuracy:.4f}")
+    print(f"最终测试集的准确率: {final_test_accuracy:.4f}")  # 打印测试集准确率