From 34fb252cc5cf4dd3e1fa1921296dbb20bd9aa031 Mon Sep 17 00:00:00 2001
From: butterfly-is-butter-fly <3425654530@qq.com>
Date: Wed, 16 Oct 2024 00:18:43 +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 | 76 ++++++++++++++++++++---------------------------------
 1 file changed, 28 insertions(+), 48 deletions(-)

diff --git a/function.py b/function.py
index 22374b0..7ac31ff 100644
--- a/function.py
+++ b/function.py
@@ -36,27 +36,33 @@ name = "邓杰元"
 - int: 预测的分类标签。
 """
 
-def normalize(dataSet):
+def mahalanobis(inX, dataSet):
     """
-    对数据进行归一化处理，使每个特征都在相同的范围内。
+    计算输入样本与数据集中每个样本的马氏距离。
+    
     Args:
-    - dataSet (numpy array): 原始数据集。
+    - inX (numpy array): 输入待分类的样本。
+    - dataSet (numpy array): 训练数据集。
 
     Returns:
-    - normDataSet (numpy array): 归一化后的数据集。
-    - ranges (numpy array): 每个特征的范围。
-    - minVals (numpy array): 每个特征的最小值。
+    - distances (numpy array): 输入样本与数据集中每个样本的马氏距离。
     """
-    minVals = dataSet.min(0)
-    maxVals = dataSet.max(0)
-    ranges = maxVals - minVals
-    normDataSet = (dataSet - minVals) / ranges
-    return normDataSet, ranges, minVals
-
+    # 计算协方差矩阵并对其求逆
+    cov_matrix = np.cov(dataSet, rowvar=False)
+    cov_matrix_inv = np.linalg.inv(cov_matrix)
+    
+    # 计算马氏距离
+    distances = []
+    for data_point in dataSet:
+        diff = data_point - inX
+        distance = np.sqrt(diff.T @ cov_matrix_inv @ diff)
+        distances.append(distance)
+    
+    return np.array(distances)
 
 def classify0(inX, dataSet, labels, k=20):
     """
-    KNN 分类算法。
+    KNN 分类算法，使用马氏距离。
 
     Args:
     - inX (numpy array): 输入待分类的样本。
@@ -67,8 +73,8 @@ def classify0(inX, dataSet, labels, k=20):
     Returns:
     - int: 预测的分类标签。
     """
-    # 计算输入向量与训练集中每个样本的欧氏距离
-    distances = np.sqrt(np.sum((dataSet - inX) ** 2, axis=1))
+    # 计算输入向量与训练集中每个样本的马氏距离
+    distances = mahalanobis(inX, dataSet)
     
     # 获取距离最小的 k 个样本的索引
     k_indices = np.argsort(distances)[:k]
@@ -79,23 +85,13 @@ def classify0(inX, dataSet, labels, k=20):
     # 手动实现一个计数器，统计每个标签出现的次数
     label_count = {}
     for label in k_nearest_labels:
-        if label in label_count:
-            label_count[label] += 1
-        else:
-            label_count[label] = 1
+        label_count[label] = label_count.get(label, 0) + 1
     
     # 找出出现次数最多的标签
-    max_count = -1
-    most_common_label = None
-    for label, count in label_count.items():
-        if count > max_count:
-            max_count = count
-            most_common_label = label
+    most_common_label = max(label_count, key=label_count.get)
     
-    # 返回出现次数最多的标签
     return most_common_label
 
-
 def load_data(file_path):
     """
     加载 TSV 格式数据文件。
@@ -112,36 +108,20 @@ def load_data(file_path):
     labels = data[:, -1].astype(int)
     return features, labels
 
-
-
 if __name__ == "__main__":
     # 测试代码，可以自行修改和测试
     
     train_file = "datingTrainSet2.tsv"
-  
     test_file = "datingTestSet2.tsv"
     
     train_data, train_labels = load_data(train_file)
     test_data, test_labels = load_data(test_file)
-    
-    # 对训练集和测试集进行归一化
-    norm_train_data, ranges, minVals = normalize(train_data)
-    norm_test_data = (test_data - minVals) / ranges
-    
-    errors = 0  # 用于统计错误分类的样本数量
+    a = 0
   
     # 对测试集中的样本进行预测
-    for i in range(len(norm_test_data)):
-        predicted_label = classify0(norm_test_data[i], norm_train_data, train_labels, k=20)
+    for i in range(len(test_data)):
+        predicted_label = classify0(test_data[i], train_data, train_labels, k=20)
         print(f"样本 {i+1}: 预测标签: {predicted_label}, 实际标签: {test_labels[i]}")
         if predicted_label != test_labels[i]:
-            errors += 1
-    
-    # 输出错误率
-    error_rate = errors / len(test_data)
-    print(f"错误分类的样本数: {errors}")
-    print(f"错误率: {error_rate * 100:.2f}%")
-
-
-
-
+            a += 1
+    print(f'错误分类的样本数: {a}')
\ No newline at end of file