Python代码实现差分隐私下的商品购买频率统计及敏感度分析

下面是一个示例代码，用于获取txt文件中的数据并进行统计。\n\npython\nimport numpy as np\nimport random\n\ndef load_data(file_path):\n data = []\n with open(file_path, 'r') as file:\n for line in file:\n line = line.strip()\n if line:\n user_items = [int(item) for item in line.split()]\n data.append(user_items)\n return data\n\ndef count_item_frequency(data):\n item_frequency = {}\n for user_items in data:\n for item in user_items:\n if item in item_frequency:\n item_frequency[item] += 1\n else:\n item_frequency[item] = 1\n return item_frequency\n\ndef add_laplace_noise(value, epsilon):\n sensitivity = 1.0\n scale = sensitivity / epsilon\n noise = np.random.laplace(0, scale)\n return value + noise\n\ndef calculate_mse(true_values, noisy_values):\n mse = np.mean((true_values - noisy_values) ** 2)\n return mse\n\n# 加载数据\ndata = load_data('data.txt')\n\n# 统计商品购买频率\nitem_frequency = count_item_frequency(data)\n\n# 添加拉普拉斯噪声\nepsilon = 1.0\nnoisy_item_frequency = {}\nfor item, frequency in item_frequency.items():\n noisy_frequency = add_laplace_noise(frequency, epsilon)\n noisy_item_frequency[item] = noisy_frequency\n\n# 计算精度（MSE）\ntrue_values = np.array(list(item_frequency.values()))\nnoisy_values = np.array(list(noisy_item_frequency.values()))\nmse = calculate_mse(true_values, noisy_values)\n\nprint("True values:", true_values)\nprint("Noisy values:", noisy_values)\nprint("MSE:", mse)\n\n\n敏感度分析可以通过在计算MSE时人为地增加一个扰动项来实现，例如：\n\npython\nsensitivity = 2.0 # 敏感度增加2\nmse_with_perturbation = calculate_mse(true_values, noisy_values + sensitivity)\nprint("MSE with perturbation:", mse_with_perturbation)\n\n\n可以根据需要调整代码中的路径、隐私预算和敏感度。