更新 function.py

function.py

@@ -40,9 +40,10 @@ name = "梅晓东"
 """
 
 
-# 这里写自己的代码实现
+import numpy as np
 import operator
-def classify0(inX, dataSet, labels, k=0):
+
+def classify0(inX, dataSet, labels, k=3):
     """
     KNN 分类算法。
 
@@ -55,10 +56,18 @@ def classify0(inX, dataSet, labels, k=0):
     Returns:
     - int: 预测的分类标签。
     """
-    dataSetSize = dataSet.shape[0]
     
+    # 确保 k 不大于训练集的大小
+    if k > dataSet.shape[0]:
+        raise ValueError(f"k={k} cannot be greater than the number of training samples {dataSet.shape[0]}.")
+
     # 计算协方差矩阵及其逆矩阵
     cov_matrix = np.cov(dataSet, rowvar=False)
+    
+    # 防止协方差矩阵不可逆
+    if np.linalg.det(cov_matrix) == 0:
+        raise ValueError("Covariance matrix is singular and cannot be inverted.")
+
     inv_cov_matrix = np.linalg.inv(cov_matrix)
     
     # 计算马氏距离
@@ -67,13 +76,20 @@ def classify0(inX, dataSet, labels, k=0):
     mahalanobis_distances = np.sqrt(np.sum(left_term * diffMat, axis=1))
 
     # 排序距离并选择前 k 个最近邻
-    sortedDistIndicies = mahalanobis_distances.argsort()
+    sortedDistIndices = mahalanobis_distances.argsort()
+    
     classCount = {}
     for i in range(k):
-        vote_label = labels[sortedDistIndicies[i]]
+        vote_label = labels[sortedDistIndices[i]]
         classCount[vote_label] = classCount.get(vote_label, 0) + 1
 
+    # 将类计数排序
     sortedClassCount = sorted(classCount.items(), key=operator.itemgetter(1), reverse=True)
+    
+    # 如果没有找到任何类别，抛出错误
+    if not sortedClassCount:
+        raise ValueError("No class votes found. Please check your training data.")
+
     return sortedClassCount[0][0]