更新 function.py

2024-10-08 23:02:38 +08:00
parent 8b948cdab4
commit 47cc47bc65
1 changed files with 64 additions and 90 deletions
@@ -40,94 +40,68 @@ name = "章程鸿"


 # 这里写自己的代码实现
-import numpy as np
-from sklearn.neighbors import KNeighborsClassifier
-from sklearn.model_selection import GridSearchCV
-
-def mahalanobis_distance(x, y, cov_inv):
-    """
-    计算两个向量之间的马氏距离。
-    
-    Args:
-    - x (numpy array): 向量1。
-    - y (numpy array): 向量2。
-    - cov_inv (numpy array): 协方差矩阵的逆矩阵。
-    
-    Returns:
-    - float: 马氏距离。
-    """
-    diff = x - y
-    return np.sqrt(np.dot(np.dot(diff, cov_inv), diff.T))
-
-def load_data(file_path):
-    """
-    加载 TSV 格式数据文件。
-    
-    Args:
-    - file_path (str): 数据文件路径。
-    
-    Returns:
-    - tuple: 特征和标签。
-    """
-    data = np.loadtxt(file_path, delimiter="\t")
-    features = data[:, :-1]
-    labels = data[:, -1].astype(int)
-    return features, labels
-
-def normalize_min_max(data):
-    """
-    对数据进行最小-最大缩放归一化。
-    
-    Args:
-    - data (numpy array): 待归一化的数据。
-    
-    Returns:
-    - tuple: 归一化后的数据、最小值、最大值。
-    """
-    min_vals = data.min(0)
-    max_vals = data.max(0)
-    ranges = max_vals - min_vals
-    # 避免除以零的情况
-    ranges[ranges == 0.0] = 1.0
-    norm_data = (data - min_vals) / ranges
-    return norm_data, min_vals, max_vals
-
-if __name__ == "__main__":
-    # 以下代码可用于测试，可自行随意修改
-    train_file = "datingTrainSet2.tsv"
-    
-    # 加载训练数据
-    train_data, train_labels = load_data(train_file)
-
-    # 归一化训练数据，并获取最小值和最大值
-    train_data_normalized, min_vals, max_vals = normalize_min_max(train_data)
-
-    # 计算协方差矩阵及其逆矩阵
-    cov_matrix = np.cov(train_data_normalized, rowvar=False)
-    inv_cov_matrix = np.linalg.inv(cov_matrix)
-
-    # 定义KNN分类器，使用自定义的马氏距离
-    knn = KNeighborsClassifier(metric=lambda x, y: mahalanobis_distance(x, y, inv_cov_matrix))
-
-    # 定义要搜索的k值范围
-    param_grid = {'n_neighbors': list(range(1, 31))}
-
-    # 使用GridSearchCV进行网格搜索
-    grid_search = GridSearchCV(knn, param_grid, cv=10, scoring='accuracy')
-    grid_search.fit(train_data_normalized, train_labels)
-
-    # 获取最佳的k值
-    optimal_k = grid_search.best_params_['n_neighbors']
-    print("Optimal k:", optimal_k)
-
-    # 使用最佳k值进行预测
-    best_knn = KNeighborsClassifier(n_neighbors=optimal_k, metric=lambda x, y: mahalanobis_distance(x, y, inv_cov_matrix))
-    best_knn.fit(train_data_normalized, train_labels)
-
-    # 测试数据点也需要用同样的方式归一化
-    test_point = np.array([65915, 3.868359, 0.128078])
-    test_point_normalized = (test_point - min_vals) / (max_vals - min_vals)
-
-    # 使用归一化后的数据进行预测
-    predicted_label = best_knn.predict([test_point_normalized])
+import numpy as np  
+from sklearn.model_selection import GridSearchCV  
+from sklearn.neighbors import KNeighborsClassifier  
+  
+def load_data(file_path):  
+    """  
+    加载 TSV 格式数据文件。  
+    """  
+    data = np.loadtxt(file_path, delimiter="\t")  
+    features = data[:, :-1]  
+    labels = data[:, -1].astype(int)  
+    return features, labels  
+  
+def normalize_min_max(data):  
+    """  
+    对数据进行最小-最大缩放归一化。  
+  
+    Args:  
+    - data (numpy array): 待归一化的数据。  
+  
+    Returns:  
+    - numpy array: 归一化后的数据。  
+    - numpy array: 最小值。  
+    - numpy array: 最大值。  
+    """  
+    min_vals = data.min(0)  
+    max_vals = data.max(0)  
+    ranges = max_vals - min_vals  
+    ranges[ranges == 0.0] = 1.0  # 避免除以零  
+    norm_data = (data - min_vals) / ranges  
+    return norm_data, min_vals, max_vals  
+  
+if __name__ == "__main__":  
+    train_file = "datingTrainSet2.tsv"  
+    test_file = "datingTestSet2.tsv"  # 假设您有一个测试集文件  
+      
+    # 加载训练和测试数据  
+    train_data, train_labels = load_data(train_file)  
+    test_data, test_labels = load_data(test_file)  
+  
+    # 归一化训练和测试数据  
+    train_data_normalized, min_vals, max_vals = normalize_min_max(train_data)  
+    test_data_normalized = (test_data - min_vals) / (max_vals - min_vals)  
+  
+    # 定义KNN分类器  
+    knn = KNeighborsClassifier()  
+  
+    # 定义要搜索的k值范围  
+    param_grid = {'n_neighbors': list(range(1, 31))}  
+  
+    # 使用GridSearchCV进行网格搜索  
+    grid_search = GridSearchCV(knn, param_grid, cv=10, scoring='accuracy')  
+    grid_search.fit(train_data_normalized, train_labels)  
+  
+    # 获取最佳的k值  
+    optimal_k = grid_search.best_params_['n_neighbors']  
+    print("Optimal k:", optimal_k)  
+  
+    # 使用最佳k值进行预测  
+    best_knn = KNeighborsClassifier(n_neighbors=optimal_k)  
+    best_knn.fit(train_data_normalized, train_labels)  
+  
+    # 对测试数据进行预测  
+    predicted_labels = best_knn.predict(test_data_normalized)  
    print("预测的分类标签是:", predicted_label[0])