generated from LufsX/Experiment-kNN
This commit is contained in:
+90
-68
@@ -41,48 +41,44 @@ name = "章程鸿·"
|
||||
|
||||
|
||||
# 这里写自己的代码实现
|
||||
import numpy as np
|
||||
import os
|
||||
from collections import Counter
|
||||
import numpy as np
|
||||
import os
|
||||
from collections import Counter
|
||||
|
||||
def classify(input_vector, dataSet, labels, k=3):
|
||||
"""
|
||||
使用 KNN 算法对输入样本进行分类。
|
||||
|
||||
参数:
|
||||
- input_vector (numpy array): 输入待分类的样本(1xN)。
|
||||
- dataSet (numpy array): 训练数据集(MxN)。
|
||||
- labels (list or numpy array): 训练集对应的标签。
|
||||
- k (int): kNN 算法参数,默认值为 3。
|
||||
|
||||
返回:
|
||||
- int: 预测的分类标签。
|
||||
"""
|
||||
# 计算马氏距离
|
||||
mean_vector = np.mean(dataSet, axis=0)
|
||||
cov_matrix = np.cov(dataSet.T)
|
||||
inv_cov_matrix = np.linalg.inv(cov_matrix)
|
||||
diff_vector = input_vector - dataSet
|
||||
mahalanobis_distances = np.sqrt(np.einsum('ij,jk,ik->i', diff_vector, inv_cov_matrix, diff_vector.T))
|
||||
sorted_indices = mahalanobis_distances.argsort()
|
||||
k_nearest_distances = mahalanobis_distances[sorted_indices[:k]]
|
||||
|
||||
# 这里写自己的代码实现
|
||||
def classify(input_vector, dataSet, labels, k=2):
|
||||
"""
|
||||
使用 KNN 算法对输入样本进行分类。
|
||||
|
||||
参数:
|
||||
- input_vector (numpy array): 输入待分类的样本(1xN)。
|
||||
- dataSet (numpy array): 训练数据集(MxN)。
|
||||
- labels (list or numpy array): 训练集对应的标签。
|
||||
- k (int): kNN 算法参数,默认值为 2。
|
||||
|
||||
返回:
|
||||
- int: 预测的分类标签。
|
||||
"""
|
||||
# 在 classify 函数中实现权重投票
|
||||
distances = np.sum((input_vector - dataSet) ** 2, axis=1)
|
||||
sorted_indices = distances.argsort()
|
||||
k_nearest_distances = distances[sorted_indices[:k]]
|
||||
k_nearest_labels = labels[sorted_indices[:k]]
|
||||
|
||||
|
||||
# 计算权重(距离的倒数)
|
||||
weights = 1 / k_nearest_distances
|
||||
# 归一化权重
|
||||
weights /= weights.sum()
|
||||
|
||||
# 使用加权多数表决法
|
||||
weighted_votes = {}
|
||||
|
||||
# 计算加权投票
|
||||
weighted_votes = np.zeros((10,))
|
||||
for label, weight in zip(k_nearest_labels, weights):
|
||||
if label not in weighted_votes:
|
||||
weighted_votes[label] = weight
|
||||
else:
|
||||
weighted_votes[label] += weight
|
||||
|
||||
return max(weighted_votes, key=weighted_votes.get)
|
||||
weighted_votes[label] += weight
|
||||
|
||||
return np.argmax(weighted_votes)
|
||||
|
||||
# 在 load_dataset 函数中添加特征缩放
|
||||
def load_dataset(directory):
|
||||
feature_list = []
|
||||
label_list = []
|
||||
@@ -93,31 +89,72 @@ def load_dataset(directory):
|
||||
label_list.append(int(file.split('_')[0]))
|
||||
feature_matrix = np.array(feature_list)
|
||||
label_array = np.array(label_list)
|
||||
|
||||
# 特征缩放(归一化)
|
||||
min_value = np.min(feature_matrix)
|
||||
max_value = np.max(feature_matrix)
|
||||
feature_matrix = (feature_matrix - min_value) / (max_value - min_value)
|
||||
|
||||
|
||||
# 特征缩放
|
||||
feature_matrix = feature_matrix / 255.0
|
||||
return feature_matrix, label_array
|
||||
# 计算输入向量与训练集之间的欧氏距离
|
||||
distances = np.sum((input_vector - dataSet) ** 2, axis=1) ** 0.5
|
||||
# 根据距离排序并获取索引
|
||||
sorted_indices = distances.argsort()
|
||||
# 选择距离最近的k个点
|
||||
k_nearest_labels = labels[sorted_indices[:k]].tolist() # 将 NumPy 数组转换为列表
|
||||
# 统计k个点中各类别的数量
|
||||
count = Counter(k_nearest_labels)
|
||||
# 返回数量最多的类别标签
|
||||
return count.most_common(1)[0][0]
|
||||
|
||||
def img_to_vector(filename):
|
||||
"""
|
||||
将二值图像转换为向量。
|
||||
|
||||
Args:
|
||||
- filename (str): 图像文件路径。
|
||||
|
||||
Returns:
|
||||
- np.array: 二值图像向量。
|
||||
"""
|
||||
vector = np.zeros((1024,))
|
||||
# 这里也写自己的代码实现
|
||||
def load_dataset(directory):
|
||||
"""
|
||||
加载数据集中的图像,并将其转换为特征向量和标签。
|
||||
|
||||
参数:
|
||||
- directory (str): 数据文件夹路径。
|
||||
|
||||
返回:
|
||||
- np.array: 特征数据(Mx1024)。
|
||||
- np.array: 标签数据(Mx1)。
|
||||
"""
|
||||
# 初始化特征数据列表和标签列表
|
||||
feature_list = []
|
||||
label_list = []
|
||||
# 遍历目录中的所有文件
|
||||
for file in os.listdir(directory):
|
||||
# 获取文件的完整路径
|
||||
file_path = os.path.join(directory, file)
|
||||
# 将图像文件转换为向量
|
||||
feature_vector = img_to_vector(file_path)
|
||||
# 将特征向量添加到列表中
|
||||
feature_list.append(feature_vector)
|
||||
# 获取文件名中的标签并添加到标签列表中
|
||||
label_list.append(int(file.split('_')[0]))
|
||||
# 将特征列表和标签列表转换为numpy数组
|
||||
feature_matrix = np.array(feature_list)
|
||||
label_array = np.array(label_list)
|
||||
return feature_matrix, label_array
|
||||
# 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。
|
||||
|
||||
Args:
|
||||
- filename (str): 图像文件路径。
|
||||
|
||||
Returns:
|
||||
- np.array: 二值图像向量。
|
||||
"""
|
||||
# 初始化向量
|
||||
vector = np.zeros((1024,)) # 创建一个一维数组
|
||||
# 打开文件并逐行读取
|
||||
with open(filename) as file:
|
||||
for i, line in enumerate(file):
|
||||
# 将每行的数字字符转换为整数并存储到向量中
|
||||
vector[i*32:i*32+32] = np.array([int(char) for char in line.strip()])
|
||||
return vector
|
||||
|
||||
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")
|
||||
|
||||
test_data = (
|
||||
@@ -154,25 +191,10 @@ if __name__ == "__main__":
|
||||
"00000000000001111111110000000000"
|
||||
"00000000000000000111100000000000"
|
||||
)
|
||||
# 使用多数表决法的结果
|
||||
predicted_label_majority = classify(
|
||||
predicted_label = classify(
|
||||
np.array([int(char) for char in test_data]).reshape(1, 1024),
|
||||
train_data,
|
||||
train_labels,
|
||||
)
|
||||
print(f"使用多数表决法预测的标签: {predicted_label_majority}")
|
||||
|
||||
# 使用加权多数表决法的结果
|
||||
predicted_label_weighted = classify(
|
||||
np.array([int(char) for char in test_data]).reshape(1, 1024),
|
||||
train_data,
|
||||
train_labels,
|
||||
)
|
||||
print(f"使用加权多数表决法预测的标签: {predicted_label_weighted}")
|
||||
|
||||
# 对比分析
|
||||
if predicted_label_majority == predicted_label_weighted:
|
||||
print("两种方法预测结果一致。")
|
||||
else:
|
||||
print("两种方法预测结果不同。")
|
||||
|
||||
print(f"预测的标签: {predicted_label}")
|
||||
Reference in New Issue
Block a user