From 2cf474c644b106801a1716e24f5e13f2d3960218 Mon Sep 17 00:00:00 2001
From: heart20003 <1789588961@qq.com>
Date: Thu, 17 Oct 2024 16:39:06 +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 | 27 +++++++++++++++------------
 1 file changed, 15 insertions(+), 12 deletions(-)

diff --git a/function.py b/function.py
index 73c6ece..6a2631e 100644
--- a/function.py
+++ b/function.py
@@ -54,25 +54,28 @@ def classify0(inX, dataSet,labels, k=3):
     inv_S[nonzero_indices] = 1 / S[nonzero_indices]
     inv_cov_matrix = VT.T @ np.diag(inv_S) @ U.T
 
-    # 计算马氏距离的平方
+    # 计算马氏距离
     diff = normalized_dataSet - normalized_inX
-    squared_distances = np.einsum('ij,jk->i', diff, inv_cov_matrix @ diff.T)
+    distances = np.sum(diff @ inv_cov_matrix * diff, axis=1)
 
     # 获取距离最小的 k 个样本的索引
-    sorted_indices = np.argsort(distances)[:k] 
+    sorted_indices = np.argsort(distances)[:k]
 
-    # 获取 k 个最近邻的标签
+    # 获取 k 个最近邻的标签和对应的距离
     nearest_labels = [labels[i] for i in sorted_indices]
+    nearest_distances = distances[sorted_indices]
 
-    # 统计 k 个最近邻中每个类别的出现次数
-    label_count = {}
-    for label in nearest_labels:
-        if label in label_count:
-            label_count[label] += 1
+    # 计算每个标签的加权投票
+    label_weight = {}
+    for i, label in enumerate(nearest_labels):
+        # 反距离作为权重，避免距离为0的情况
+        weight = 1 / (nearest_distances[i] + 1e-10)  # 加一个小值避免除以零
+        if label in label_weight:
+            label_weight[label] += weight
         else:
-            label_count[label] = 1
+            label_weight[label] = weight
 
-    # 找到出现次数最多的标签
-    predicted_label = max(label_count, key=label_count.get)
+    # 找到加权投票中出现次数最多的标签
+    predicted_label = max(label_weight, key=label_weight.get)
 
     return predicted_label