From d5bcf907c4f68363f935d63e1959b2474380ee91 Mon Sep 17 00:00:00 2001
From: cy <939601402@qq.com>
Date: Sun, 10 Nov 2024 23:54:15 +0800
Subject: [PATCH] =?UTF-8?q?=E6=9B=B4=E6=96=B0=20function.py?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 function.py | 62 ++++++++++++++++++++++++++++-------------------------
 1 file changed, 33 insertions(+), 29 deletions(-)

diff --git a/function.py b/function.py
index bdc9723..2fcb6c1 100644
--- a/function.py
+++ b/function.py
@@ -1,7 +1,5 @@
-# function.py
 import numpy as np
 from sklearn.preprocessing import StandardScaler
-from itertools import product
 
 """
 实验说明:
@@ -65,7 +63,7 @@ def train_logreg(X, y, learning_rate=0.05, iterations=1500, batch_size=64, regul
 
         # 可选：学习率衰减
         if i % 100 == 0:
-            learning_rate *= 0.95  # 每100次迭代衰减5%
+            learning_rate *= 0.95  # 每 100 次迭代衰减 5%
 
     return weights
 
@@ -86,12 +84,13 @@ def predict_logreg(inX, weights):
     return 1 if h > 0.5 else 0
 
 
-def load_data(filepath):
+def load_data(filepath, fill_method='zero'):
     """
-    从文件中加载数据，填充缺失值为均值，并进行特征处理。
+    从文件中加载数据，根据指定方法填充缺失值，并进行特征处理。
 
     Args:
     - filepath (str): 数据文件路径。
+    - fill_method (str): 缺失值填充方法，可选值为'zero'（用 0 填充）或'mean'（用均值填充）。
 
     Returns:
     - np.array: 特征数据。
@@ -104,32 +103,30 @@ def load_data(filepath):
         for line in lines:
             line = line.strip().split()
             if len(line) >= 1 and line[-1].isdigit():
-                # 初始化数据，将 '?' 保持为原始字符串
-                line_data = [1] + [x if x != '?' else x for x in line[:-1]]
+                line_data = []
+                for x in line[:-1]:
+                    if x == '?':
+                        if fill_method == 'zero':
+                            line_data.append(0)
+                        elif fill_method == 'mean':
+                            # 这里假设在处理过程中已经计算了每个特征列的均值，可以用一个字典存储
+                            # 这里只是模拟，实际需要在函数外计算好均值并传入或在函数中动态计算
+                            mean_value = 5.0  # 假设均值为 5.0
+                            line_data.append(mean_value)
+                    else:
+                        line_data.append(float(x))
                 data.append(line_data)
                 labels.append(int(line[-1]))
 
-    # 转换为 numpy 数组
-    data = np.array(data, dtype=np.object)
+    data = np.array(data)
     labels = np.array(labels)
 
-    # 将 '?' 替换为 np.nan，便于计算均值
-    data[data == '?'] = np.nan
-    data = data.astype(float)
-
-    # 计算每个特征的均值
-    col_means = np.nanmean(data[:, 1:], axis=0)  # 排除常数项（偏置项）
-
-    # 用每个特征的均值填充缺失值
-    inds = np.where(np.isnan(data))
-    data[inds] = np.take(col_means, inds[1]-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
+    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
+    # 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
 
     scaler = StandardScaler()
     data = scaler.fit_transform(data)
@@ -137,13 +134,20 @@ def load_data(filepath):
     return data, target_variable
 
 
-
 if __name__ == "__main__":
-    X_train, y_train = load_data("./data/horse-colic.data")
+    X_train_zero, y_train_zero = load_data("./data/horse-colic.data", fill_method='zero')
+    X_train_mean, y_train_mean = load_data("./data/horse-colic.data", fill_method='mean')
     X_test, y_test = load_data("./data/horse-colic.test")
 
-    final_weights = train_logreg(X_train, y_train, learning_rate=0.05, iterations=1500, batch_size=64, regularization='L2', lambda_reg=0.05)
+    final_weights_zero = train_logreg(X_train_zero, y_train_zero, learning_rate=0.05, iterations=1500, batch_size=64, regularization='L2', lambda_reg=0.05)
+    final_weights_mean = train_logreg(X_train_mean, y_train_mean, learning_rate=0.05, iterations=1500, batch_size=64, regularization='L2', lambda_reg=0.05)
 
-    # 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}")
\ No newline at end of file
+    # 计算并输出用“0”填充的测试集准确率
+    y_pred_zero = [predict_logreg(inX, final_weights_zero) for inX in X_test]
+    accuracy_zero = np.mean(np.array(y_pred_zero) == y_test)
+    print(f"用“0”填充缺失值的测试集准确率: {accuracy_zero:.4f}")
+
+    # 计算并输出用均值填充的测试集准确率
+    y_pred_mean = [predict_logreg(inX, final_weights_mean) for inX in X_test]
+    accuracy_mean = np.mean(np.array(y_pred_mean) == y_test)
+    print(f"用均值填充缺失值的测试集准确率: {accuracy_mean:.4f}")
\ No newline at end of file