generated from LufsX/Experiment-kNN
This commit is contained in:
lzh
+9
-128
@@ -41,83 +41,10 @@ name = "廖子豪"
|
||||
|
||||
|
||||
# 这里写自己的代码实现
|
||||
def classify(input_vector, dataSet, labels, k_values=[1,31]):
|
||||
"""
|
||||
KNN 加权投票分类算法，针对多个k值进行预测，并选出不同k值下得票最多的类别。
|
||||
|
||||
Args:
|
||||
- input_vector (numpy array): 输入待分类的样本。
|
||||
- dataSet (numpy array): 训练数据集。
|
||||
- labels (list or numpy array): 训练集对应的标签。
|
||||
- k_values (list of int): 多个kNN 算法的k值列表。
|
||||
|
||||
Returns:
|
||||
- int: 预测的分类标签（基于多个k值投票的结果）。
|
||||
"""
|
||||
num_samples = dataSet.shape[0] # 读取样本数量
|
||||
|
||||
# step 1: 计算欧式距离
|
||||
diff = np.tile(input_vector, (num_samples, 1)) - dataSet # 元素逐个相减
|
||||
squared_diff = diff ** 2 # 平方
|
||||
squared_dist = np.sum(squared_diff, axis=1) # 按行求和
|
||||
distance = squared_dist ** 0.5 # 计算欧式距离
|
||||
|
||||
# step 2: 对距离进行排序，返回排序后的索引
|
||||
sorted_dist_indices = np.argsort(distance)
|
||||
|
||||
# 保存每个k值对应的预测结果
|
||||
vote_results = []
|
||||
|
||||
for k in k_values:
|
||||
class_count = {} # 定义一个空字典
|
||||
|
||||
for i in range(k):
|
||||
# step 3: 选择距离最小的k个样本
|
||||
vote_label = labels[sorted_dist_indices[i]]
|
||||
dist = distance[sorted_dist_indices[i]]
|
||||
|
||||
# step 4: 使用距离倒数作为权重（避免除0，设置一个最小距离）
|
||||
weight = 1 / (dist + 1e-5) # 添加一个小常数避免除以0
|
||||
|
||||
# step 5: 进行加权投票
|
||||
class_count[vote_label] = class_count.get(vote_label, 0) + weight
|
||||
|
||||
# step 6: 找到加权后得票最多的类别作为该k值的预测结果
|
||||
max_vote = max(class_count, key=class_count.get)
|
||||
vote_results.append(max_vote)
|
||||
|
||||
# step 7: 对所有k值的预测结果进行投票，统计出现次数最多的标签
|
||||
final_vote_count = {}
|
||||
for result in vote_results:
|
||||
final_vote_count[result] = final_vote_count.get(result, 0) + 1
|
||||
|
||||
# step 8: 返回得票最多的类别作为最终分类结果
|
||||
final_result = max(final_vote_count, key=final_vote_count.get)
|
||||
return final_result
|
||||
|
||||
|
||||
# 这里也写自己的代码实现
|
||||
# ... 你的代码 ...
|
||||
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。
|
||||
"""
|
||||
"""
|
||||
从文件夹中加载数据。
|
||||
|
||||
Args:
|
||||
- directory (str): 数据文件夹路径。
|
||||
|
||||
Returns:
|
||||
- np.array: 特征数据。
|
||||
- np.array: 标签数据。
|
||||
"""
|
||||
labels = []
|
||||
file_list = os.listdir(directory)
|
||||
num_samples = len(file_list)
|
||||
@@ -130,66 +57,20 @@ def load_dataset(directory):
|
||||
|
||||
return dataset, np.array(labels)
|
||||
|
||||
# 这里还写自己的代码实现
|
||||
def img_to_vector(filename):
|
||||
"""
|
||||
将二值图像转换为向量。
|
||||
该部分应为 load_dataset 函数的辅助子函数，可单独在这里实现，也可直接集成至 load_dataset 函数中。
|
||||
|
||||
Args:
|
||||
- filename (str): 图像文件路径。
|
||||
|
||||
Returns:
|
||||
- np.array: 二值图像向量。
|
||||
"""
|
||||
rows = 32
|
||||
columns = 32
|
||||
img_vector = np.zeros((1, rows * columns))
|
||||
file_in = open(filename)
|
||||
for row in range(rows):
|
||||
line_str = file_in.readline()
|
||||
for col in range(columns):
|
||||
img_vector[0, row * 32 + col] = int(line_str[col])
|
||||
|
||||
with open(filename, 'r') as file_in:
|
||||
img_vector = np.zeros((1, 1024))
|
||||
for idx, line in enumerate(file_in):
|
||||
for col_idx, char in enumerate(line.strip()):
|
||||
img_vector[0, idx * 32 + col_idx] = int(char)
|
||||
return img_vector
|
||||
|
||||
if __name__ == "__main__":
|
||||
# 以下代码可用于测试，可自行随意修改
|
||||
# 请确保trainingDigits文件夹在你的工作目录中，或者提供完整的路径
|
||||
train_data, train_labels = load_dataset("trainingDigits")
|
||||
|
||||
test_data = (
|
||||
"00000000000000110000000000000000"
|
||||
"00000000000011111000000000000000"
|
||||
"00000000000011111100000000000000"
|
||||
"00000000001111111110000000000000"
|
||||
"00000000001111111110000000000000"
|
||||
"00000000001111111100000000000000"
|
||||
"00000000011111111000000000000000"
|
||||
"00000000111111111000000000000000"
|
||||
"00000000111111110000000000000000"
|
||||
"00000000111111110000000000000000"
|
||||
"00000000111111110000000000000000"
|
||||
"00000000111111110000000000000000"
|
||||
"00000000111111100000000000000000"
|
||||
"00000001111111100000000000000000"
|
||||
"00000001111111100000000000000000"
|
||||
"00000001111111100000000000000000"
|
||||
"00000001111111111111000000000000"
|
||||
"00000001111111111111111000000000"
|
||||
"00000001111111111111111100000000"
|
||||
"00000011111111111111111110000000"
|
||||
"00000011111111111111111110000000"
|
||||
"00000001111111111111111110000000"
|
||||
"00000001111111111111111110000000"
|
||||
"00000000111111111100111111100000"
|
||||
"00000000111111111000111111100000"
|
||||
"00000000011111111001111111100000"
|
||||
"00000000011111111111111111000000"
|
||||
"00000000001111111111111111000000"
|
||||
"00000000000111111111111111000000"
|
||||
"00000000000011111111111100000000"
|
||||
"00000000000001111111110000000000"
|
||||
"00000000000000000111100000000000"
|
||||
# ... 你的测试数据 ...
|
||||
)
|
||||
predicted_label = classify(
|
||||
np.array([int(char) for char in test_data]).reshape(1, 1024),
|
||||
@@ -197,4 +78,4 @@ if __name__ == "__main__":
|
||||
train_labels,
|
||||
)
|
||||
|
||||
print(f"预测的标签: {predicted_label}")
|
||||
print(f"预测的标签: {predicted_label}")
|
||||
Reference in New Issue
Block a user