更新 function.py

function.py

@@ -41,9 +41,9 @@ name = "黎世浩"
 
 
 # 这里写自己的代码实现
-def classify0(inX, dataSet, labels, k=2):
+def classify0(inX, dataSet, labels, k=3):
     """
-    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。
@@ -54,21 +54,21 @@ def classify0(inX, dataSet, labels, k=2):
     Returns:
     - int: 预测的分类标签。
     """
-    # 计算输入样本与数据集中每个样本的欧氏距离
+    # 计算马氏距离
-    diffMat = np.tile(inX, (dataSet.shape[0], 1)) - dataSet
+    inv_cov = np.linalg.inv(np.cov(dataSet.T))  # 计算协方差矩阵的逆
-    sqDiffMat = diffMat ** 2
+    meanVals = np.mean(dataSet, axis=0)  # 计算均值
-    sqDistances = sqDiffMat.sum(axis=1)
+    diffMat = inX - meanVals  # 计算差值
-    distances = sqDistances ** 0.5
+    mahalDist = np.sqrt(diffMat.dot(inv_cov).dot(diffMat.T))  # 计算马氏距离
-    
+
     # 对距离进行排序，选择距离最小的 k 个样本
-    sortedDistIndices = distances.argsort()
+    sortedDistIndices = np.argsort(mahalDist)
     classCount = {}
-    
+
     # 遍历距离最小的 k 个样本，统计每个类别的出现次数
     for i in range(k):
         voteIlabel = labels[sortedDistIndices[i]]
         classCount[voteIlabel] = classCount.get(voteIlabel, 0) + 1
-    
+
     # 选择出现次数最多的类别作为预测标签
     sortedClassCount = sorted(classCount.items(), key=lambda x: x[1], reverse=True)
     return sortedClassCount[0][0]
@@ -85,4 +85,4 @@ def load_data(file_path):
 if __name__ == "__main__":
     train_file = "datingTrainSet2.tsv"
     train_data, train_labels = load_data(train_file)
-    print(classify0([65915, 3.868359, 0.128078], train_data, train_labels))
+    print(classify0([65915, 3.868359, 0.128078], train_data, train_labels))