更新 function.py

function.py

@@ -41,58 +41,38 @@ name = "万子玉"
 """
 
 # 这里写自己的代码实现
-def mahalanobis_distance(x, mean, inv_covmat):
-    delta = x - mean
-    m_dist = np.sqrt(np.dot(np.dot(delta, inv_covmat), delta.T))
-    return m_dist
+def euclidean_distance(a, b):
+    return np.sqrt(np.sum((a - b) ** 2))
 
-def classify(input_vector, dataSet, labels, k=2, reg=1e-5):
-    # 确保dataSet是二维的，每个样本是一维的
-    if dataSet.ndim > 2:
-        dataSet = dataSet.reshape(dataSet.shape[0], -1)
-    
-    mean = np.mean(dataSet, axis=0)
-    covmat = np.cov(dataSet, rowvar=0)
-    
-    # 正则化协方差矩阵以避免奇异矩阵
-    covmat += np.eye(covmat.shape[0]) * reg
-    
-    inv_covmat = np.linalg.inv(covmat)
-    
-    # 马氏距离计算
-    dists = [mahalanobis_distance(input_vector, mean, inv_covmat)]
-    for data in dataSet:
-        dists.append(mahalanobis_distance(data, mean, inv_covmat))
-    
-    # 移除第一个距离，因为它是输入向量的马氏距离
-    dists.pop(0)
-    
+
+def classify(input_vector, dataSet, labels, k=2):
+    dists = [euclidean_distance(input_vector, data) for data in dataSet]
     k_indices = np.argsort(dists)[:k]
     k_nearest_labels = [labels[i] for i in k_indices]
+    k_nearest_dists = [dists[i] for i in k_indices]
     
     if not k_nearest_labels:
         raise ValueError("No neighbors found; possible causes: k is too small or all distances are the same.")
     
-    most_common = Counter(k_nearest_labels).most_common(1)
+    # 计算权重，距离越小权重越大
+    weights = [1/dist if dist != 0 else 0 for dist in k_nearest_dists]
+    total_weights = sum(weights)
+    weighted_votes = Counter()
+
+    for label, weight in zip(k_nearest_labels, weights):
+        weighted_votes[label] += weight
     
-    if most_common:
-        return most_common[0][0]
+    # 根据权重排序
+    sorted_weighted_votes = sorted(weighted_votes.items(), key=lambda x: x[1], reverse=True)
+    
+    if sorted_weighted_votes:
+        return sorted_weighted_votes[0][0]
     else:
         return None
 
 
 # 这里也写自己的代码实现
 def load_dataset(directory):
-    """
-    从文件夹中加载数据，若不自行正确实现，则无法通过评测，**不记录成绩**。
-
-    Args:
-    - directory (str): 数据文件夹路径。
-
-    Returns:
-    - np.array: 特征数据。
-    - np.array: 标签数据。
-    """
     data_list = []
     label_list = []
     for root, dirs, files in os.walk(directory):
@@ -104,6 +84,7 @@ def load_dataset(directory):
     return np.array(data_list), np.array(label_list)
 
 
+
 # 这里还写自己的代码实现
 def img_to_vector(filename):
     """
@@ -121,6 +102,7 @@ def img_to_vector(filename):
     vector = np.array([int(char) for line in lines for char in line.strip() if char.isdigit()])
     return vector.reshape(1, -1)
 
+
 if __name__ == "__main__":
     # 以下代码可用于测试，可自行随意修改
     train_data, train_labels = load_dataset("trainingDigits")
@@ -159,7 +141,7 @@ if __name__ == "__main__":
         "00000000000001111111110000000000"
         "00000000000000000111100000000000"
     )
-    predicted_label = classify(
+   predicted_label = classify(
         np.array([int(char) for char in test_data]).reshape(1, 1024),
         train_data,
         train_labels,