{"id":3895,"date":"2025-09-13T22:20:43","date_gmt":"2025-09-13T14:20:43","guid":{"rendered":"http:\/\/viplao.com\/?p=3895"},"modified":"2025-09-13T22:20:44","modified_gmt":"2025-09-13T14:20:44","slug":"%e3%80%90python%e5%ae%9e%e8%b7%b5%e6%a1%88%e4%be%8b%e3%80%91%e7%94%b5%e5%95%86%e5%b9%b3%e5%8f%b0%e6%95%b0%e6%8d%ae%e5%88%86%e6%9e%90%e5%92%8c%e6%8c%96%e6%8e%98-%e7%94%a8%e6%88%b7%e6%b5%81%e5%a4%b1","status":"publish","type":"post","link":"http:\/\/viplao.com\/index.php\/2025\/09\/13\/%e3%80%90python%e5%ae%9e%e8%b7%b5%e6%a1%88%e4%be%8b%e3%80%91%e7%94%b5%e5%95%86%e5%b9%b3%e5%8f%b0%e6%95%b0%e6%8d%ae%e5%88%86%e6%9e%90%e5%92%8c%e6%8c%96%e6%8e%98-%e7%94%a8%e6%88%b7%e6%b5%81%e5%a4%b1\/","title":{"rendered":"\u3010Python\u5b9e\u8df5\u6848\u4f8b\u3011\u7535\u5546\u5e73\u53f0\u6570\u636e\u5206\u6790\u548c\u6316\u6398 &#8211; \u7528\u6237\u6d41\u5931\u9884\u6d4b"},"content":{"rendered":"\n<h3 class=\"wp-block-heading\">\u5f00\u53d1\u601d\u8def:\u6a21\u62df\u4e865000\u540d\u7528\u6237\u7684\u6570\u636e\uff0c\u5305\u542b\u7528\u6237\u753b\u50cf\uff08\u5e74\u9f84\u3001\u6027\u522b\u3001\u4f1a\u5458\u7b49\u7ea7\u3001\u57ce\u5e02\u7ea7\u522b\uff09\u3001\u884c\u4e3a\u7279\u5f81\uff08\u8ba2\u5355\u6570\u3001\u6d88\u8d39\u989d\u3001\u8ddd\u4e0a\u6b21\u8d2d\u4e70\u5929\u6570\u3001\u8bbf\u95ee\u9891\u7387\u3001\u8d2d\u7269\u8f66\u653e\u5f03\u7387\u3001\u5ba2\u670d\u8054\u7cfb\u3001\u6298\u6263\u4f7f\u7528\uff09\u4ee5\u53ca\u4e00\u4e2a\u57fa\u4e8e\u8fd9\u4e9b\u7279\u5f81\u8ba1\u7b97\u5f97\u51fa\u7684<strong>\u6d41\u5931\u6807<\/strong>\u8fdb\u884c\u6df1\u5165\u5206\u6790\u4e0e\u6316\u6398<\/h3>\n\n\n\n<ol>\n<li><strong>\u914d\u7f6e (<code># --- \u914d\u7f6e ---<\/code>)<\/strong>:\n<ul>\n<li><code>NUM_USERS<\/code>: \u6a21\u62df\u7684\u7528\u6237\u6570\u91cf\u3002<\/li>\n\n\n\n<li><code>REPORT_PREFIX<\/code>: \u751f\u6210\u62a5\u544a\u548c\u56fe\u8868\u6587\u4ef6\u540d\u7684\u524d\u7f00\u3002<\/li>\n\n\n\n<li><code>RANDOM_SEED<\/code>: \u968f\u673a\u79cd\u5b50\uff0c\u786e\u4fdd\u7ed3\u679c\u53ef\u590d\u73b0\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li><strong>\u6570\u636e\u751f\u6210 (<code>generate_sample_churn_data<\/code>)<\/strong>:\n<ul>\n<li>\u6a21\u62df\u4e865000\u540d\u7528\u6237\u7684\u6570\u636e\uff0c\u5305\u542b\u7528\u6237\u753b\u50cf\uff08\u5e74\u9f84\u3001\u6027\u522b\u3001\u4f1a\u5458\u7b49\u7ea7\u3001\u57ce\u5e02\u7ea7\u522b\uff09\u3001\u884c\u4e3a\u7279\u5f81\uff08\u8ba2\u5355\u6570\u3001\u6d88\u8d39\u989d\u3001\u8ddd\u4e0a\u6b21\u8d2d\u4e70\u5929\u6570\u3001\u8bbf\u95ee\u9891\u7387\u3001\u8d2d\u7269\u8f66\u653e\u5f03\u7387\u3001\u5ba2\u670d\u8054\u7cfb\u3001\u6298\u6263\u4f7f\u7528\uff09\u4ee5\u53ca\u4e00\u4e2a\u57fa\u4e8e\u8fd9\u4e9b\u7279\u5f81\u8ba1\u7b97\u5f97\u51fa\u7684<strong>\u6d41\u5931\u6807\u7b7e (<code>is_churned<\/code>)<\/strong>\u3002<\/li>\n\n\n\n<li>\u6570\u636e\u7684\u751f\u6210\u903b\u8f91\u8003\u8651\u4e86\u7279\u5f81\u4e4b\u95f4\u7684\u5173\u8054\u6027\uff08\u4f8b\u5982\uff0c\u9ad8\u4ef7\u503c\u4f1a\u5458\u901a\u5e38\u66f4\u6d3b\u8dc3\uff0c\u6d41\u5931\u98ce\u9669\u66f4\u4f4e\uff09\u3002<\/li>\n\n\n\n<li>\u751f\u6210\u7684\u6570\u636e\u4fdd\u5b58\u4e3aCSV\u6587\u4ef6\uff0c\u65b9\u4fbf\u540e\u7eed\u67e5\u770b\u548c\u4f7f\u7528\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li><strong>\u6570\u636e\u9884\u5904\u7406 (<code>preprocess_data<\/code>)<\/strong>:\n<ul>\n<li>\u5bf9\u5206\u7c7b\u53d8\u91cf\uff08\u6027\u522b\u3001\u4f1a\u5458\u7b49\u7ea7\u3001\u57ce\u5e02\u7ea7\u522b\uff09\u8fdb\u884c\u6807\u7b7e\u7f16\u7801 (<code>LabelEncoder<\/code>)\uff0c\u5c06\u5176\u8f6c\u6362\u4e3a\u6570\u503c\u3002<\/li>\n\n\n\n<li>\u9009\u62e9\u7528\u4e8e\u5efa\u6a21\u7684\u7279\u5f81\u5217\u3002<\/li>\n\n\n\n<li>\u5206\u79bb\u7279\u5f81\u77e9\u9635&nbsp;<code>X<\/code>&nbsp;\u548c\u76ee\u6807\u6807\u7b7e&nbsp;<code>y<\/code>\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li><strong>\u6a21\u578b\u8bad\u7ec3\u4e0e\u8bc4\u4f30 (<code>train_and_evaluate_models<\/code>)<\/strong>:\n<ul>\n<li>\u5c06\u6570\u636e\u5212\u5206\u4e3a\u8bad\u7ec3\u96c6\u548c\u6d4b\u8bd5\u96c6\u3002<\/li>\n\n\n\n<li>\u5bf9\u7279\u5f81\u8fdb\u884c\u6807\u51c6\u5316 (<code>StandardScaler<\/code>)\uff0c\u8fd9\u5bf9\u903b\u8f91\u56de\u5f52\u6a21\u578b\u5f88\u91cd\u8981\u3002<\/li>\n\n\n\n<li>\u8bad\u7ec3\u4e24\u4e2a\u6a21\u578b\uff1a\n<ul>\n<li><strong>\u903b\u8f91\u56de\u5f52 (Logistic Regression)<\/strong>: \u7ebf\u6027\u6a21\u578b\uff0c\u7b80\u5355\u9ad8\u6548\uff0c\u53ef\u89e3\u91ca\u6027\u5f3a\u3002<\/li>\n\n\n\n<li><strong>\u968f\u673a\u68ee\u6797 (Random Forest)<\/strong>: \u975e\u7ebf\u6027\u6a21\u578b\uff0c\u901a\u5e38\u80fd\u6355\u6349\u66f4\u590d\u6742\u7684\u6a21\u5f0f\uff0c\u6027\u80fd\u53ef\u80fd\u66f4\u597d\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li>\u5728\u6d4b\u8bd5\u96c6\u4e0a\u8bc4\u4f30\u4e24\u4e2a\u6a21\u578b\u7684\u6027\u80fd\uff0c\u4f7f\u7528\u591a\u79cd\u6307\u6807\uff1a\u51c6\u786e\u7387\u3001\u7cbe\u786e\u7387\u3001\u53ec\u56de\u7387\u3001F1\u5206\u6570\u3001AUC-ROC\u3002<\/li>\n\n\n\n<li>\u6bd4\u8f83\u4e24\u4e2a\u6a21\u578b\u7684\u6027\u80fd\uff0c\u5e76\u9009\u62e9\u8868\u73b0\u6700\u597d\u7684\u4e00\u4e2a\uff08\u4ee5AUC-ROC\u4e3a\u6807\u51c6\uff09\u3002<\/li>\n\n\n\n<li>\u4e3a\u6700\u4f73\u6a21\u578b\u751f\u6210\u5e76\u4fdd\u5b58\u6df7\u6dc6\u77e9\u9635\u56fe\u8868\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li><strong>\u7279\u5f81\u91cd\u8981\u6027\u5206\u6790 (<code>analyze_feature_importance<\/code>)<\/strong>:\n<ul>\n<li>\u5206\u6790\u6700\u4f73\u6a21\u578b\u7684\u7279\u5f81\u91cd\u8981\u6027\u3002\n<ul>\n<li>\u5bf9\u4e8e\u968f\u673a\u68ee\u6797\uff0c\u4f7f\u7528\u5176\u5185\u7f6e\u7684&nbsp;<code>feature_importances_<\/code>\u3002<\/li>\n\n\n\n<li>\u5bf9\u4e8e\u903b\u8f91\u56de\u5f52\uff0c\u4f7f\u7528\u7cfb\u6570\u7684\u7edd\u5bf9\u503c&nbsp;<code>np.abs(model.coef_[0])<\/code>\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li>\u5bf9\u7279\u5f81\u6309\u91cd\u8981\u6027\u8fdb\u884c\u6392\u5e8f\uff0c\u5e76\u6253\u5370Top 10\u3002<\/li>\n\n\n\n<li>\u4f7f\u7528\u6761\u5f62\u56fe\u53ef\u89c6\u5316\u7279\u5f81\u91cd\u8981\u6027\uff0c\u5e76\u4fdd\u5b58\u56fe\u8868\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li><strong>\u62a5\u544a\u751f\u6210 (<code>generate_churn_prediction_report<\/code>)<\/strong>:\n<ul>\n<li>\u6c47\u603b\u6574\u4e2a\u5206\u6790\u8fc7\u7a0b\uff0c\u5305\u62ec\u9879\u76ee\u6982\u8ff0\u3001\u6570\u636e\u63cf\u8ff0\u3001\u6a21\u578b\u6027\u80fd\u5bf9\u6bd4\u3001\u7279\u5f81\u91cd\u8981\u6027\u5206\u6790\u3002<\/li>\n\n\n\n<li>\u63d0\u4f9b\u57fa\u4e8e\u5206\u6790\u7ed3\u679c\u7684\u5177\u4f53\u4e1a\u52a1\u5e94\u7528\u5efa\u8bae\uff0c\u4f8b\u5982\u5982\u4f55\u9488\u5bf9\u9ad8\u98ce\u9669\u7528\u6237\u8fdb\u884c\u5e72\u9884\u3002<\/li>\n\n\n\n<li>\u5c06\u6240\u6709\u5185\u5bb9\u6574\u5408\u5230\u4e00\u4e2a&nbsp;<code>.txt<\/code>&nbsp;\u6587\u672c\u62a5\u544a\u4e2d\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li><strong>\u4e3b\u51fd\u6570 (<code>main<\/code>)<\/strong>:\n<ul>\n<li>\u6309\u987a\u5e8f\u8c03\u7528\u4e0a\u8ff0\u6240\u6709\u6b65\u9aa4\u7684\u51fd\u6570\uff0c\u5b8c\u6210\u4ece\u6570\u636e\u751f\u6210\u5230\u62a5\u544a\u8f93\u51fa\u7684\u6574\u4e2a\u6d41\u7a0b\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ol>\n\n\n\n<p><\/p>\n\n\n\n<p><\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>import pandas as pd\nimport numpy as np\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.linear_model import LogisticRegression\nfrom sklearn.ensemble import RandomForestClassifier\nfrom sklearn.metrics import accuracy_score, precision_score, recall_score, f1_score, roc_auc_score, confusion_matrix, classification_report\nfrom sklearn.preprocessing import StandardScaler, LabelEncoder\nimport matplotlib.pyplot as plt\nimport seaborn as sns\nfrom datetime import datetime, timedelta\nimport warnings\nwarnings.filterwarnings(\"ignore\")\n\n# --- \u914d\u7f6e ---\nNUM_USERS = 5000\nREPORT_PREFIX = '\u7535\u5546\u7528\u6237\u6d41\u5931\u9884\u6d4b\u62a5\u544a'\nRANDOM_SEED = 42\n\n# --- \u6570\u636e\u751f\u6210 ---\n\ndef generate_sample_churn_data(n_users):\n    \"\"\"\u751f\u6210\u6a21\u62df\u7684\u7528\u6237\u6d41\u5931\u6570\u636e\"\"\"\n    print(\"--- \u6b63\u5728\u751f\u6210\u6a21\u62df\u7528\u6237\u6570\u636e ---\")\n    np.random.seed(RANDOM_SEED)\n    \n    data = &#91;]\n    user_ids = &#91;f'user_{i}' for i in range(1, n_users + 1)]\n    \n    for user_id in user_ids:\n        # --- \u7528\u6237\u57fa\u7840\u753b\u50cf ---\n        age = np.random.randint(18, 65)\n        gender = np.random.choice(&#91;'Male', 'Female'], p=&#91;0.5, 0.5])\n        membership_tier = np.random.choice(&#91;'Bronze', 'Silver', 'Gold'], p=&#91;0.6, 0.3, 0.1])\n        location_city_tier = np.random.choice(&#91;'Tier_1', 'Tier_2', 'Tier_3'], p=&#91;0.3, 0.4, 0.3])\n        \n        # --- \u7528\u6237\u884c\u4e3a\u7279\u5f81 ---\n        # \u6ce8\u518c\u65e5\u671f (\u5047\u8bbe\u57282\u5e74\u524d\u52301\u5e74\u524d\u4e4b\u95f4)\n        signup_date = datetime.now() - timedelta(days=np.random.randint(365, 2*365))\n        \n        # \u603b\u8ba2\u5355\u6570\u548c\u603b\u6d88\u8d39 (\u4e0e\u4f1a\u5458\u7b49\u7ea7\u76f8\u5173)\n        if membership_tier == 'Gold':\n            total_orders = np.random.poisson(20)\n            total_spent = np.random.lognormal(10.5, 0.4)\n        elif membership_tier == 'Silver':\n            total_orders = np.random.poisson(10)\n            total_spent = np.random.lognormal(9.5, 0.5)\n        else: # Bronze\n            total_orders = np.random.poisson(5)\n            total_spent = np.random.lognormal(8.5, 0.6)\n            \n        # \u5e73\u5747\u8ba2\u5355\u4ef7\u503c\n        avg_order_value = total_spent \/ max(total_orders, 1) \n        \n        # \u6700\u8fd1\u4e00\u6b21\u8d2d\u4e70\u8ddd\u79bb\u5929\u6570 (\u5173\u952e\u7279\u5f81)\n        # \u5047\u8bbeGold\u7528\u6237\u66f4\u6d3b\u8dc3\uff0c\u6d41\u5931\u98ce\u9669\u4f4e\uff1bBronze\u7528\u6237\u6d41\u5931\u98ce\u9669\u9ad8\n        if membership_tier == 'Gold':\n            days_since_last_purchase = np.random.exponential(15) # \u5e73\u574715\u5929\n        elif membership_tier == 'Silver':\n            days_since_last_purchase = np.random.exponential(30) # \u5e73\u574730\u5929\n        else: # Bronze\n            days_since_last_purchase = np.random.exponential(60) # \u5e73\u574760\u5929\n            \n        last_purchase_date = datetime.now() - timedelta(days=days_since_last_purchase)\n        \n        # \u6708\u5747\u8bbf\u95ee\u6b21\u6570 (\u6d3b\u8dc3\u5ea6)\n        if membership_tier == 'Gold':\n            monthly_visits = np.random.poisson(15)\n        elif membership_tier == 'Silver':\n            monthly_visits = np.random.poisson(10)\n        else: # Bronze\n            monthly_visits = np.random.poisson(5)\n            \n        # \u8d2d\u7269\u8f66\u653e\u5f03\u7387 (\u52a0\u8d2d\u4f46\u672a\u8d2d\u4e70\u7684\u8ba2\u5355 \/ \u603b\u52a0\u8d2d\u6b21\u6570)\n        cart_abandonment_rate = np.random.beta(2, 5) # \u5927\u591a\u6570\u7528\u6237\u653e\u5f03\u7387\u8f83\u4f4e\n        \n        # \u5ba2\u670d\u8054\u7cfb\u6b21\u6570 (\u6700\u8fd1\u4e00\u5e74)\n        customer_service_contacts = np.random.poisson(2)\n        \n        # \u6298\u6263\u4f7f\u7528\u9891\u7387\n        discount_usage_freq = np.random.beta(3, 7) # \u5927\u591a\u6570\u7528\u6237\u4e0d\u5e38\u7528\u6298\u6263\n        \n        # --- \u6784\u9020\u6d41\u5931\u6807\u7b7e (\u57fa\u4e8e\u7279\u5f81\u7684\u6982\u7387) ---\n        # \u8fd9\u662f\u4e00\u4e2a\u7b80\u5316\u7684\u6a21\u62df\u903b\u8f91\uff0c\u771f\u5b9e\u573a\u666f\u4f1a\u66f4\u590d\u6742\n        churn_prob = 0.0\n        \n        # \u957f\u65f6\u95f4\u672a\u8d2d\u4e70\u662f\u5f3a\u4fe1\u53f7\n        if days_since_last_purchase &gt; 90:\n            churn_prob += 0.4\n        elif days_since_last_purchase &gt; 60:\n            churn_prob += 0.2\n        elif days_since_last_purchase &gt; 30:\n            churn_prob += 0.1\n            \n        # \u8d2d\u7269\u8f66\u653e\u5f03\u7387\u9ad8\u53ef\u80fd\u8868\u793a\u72b9\u8c6b\u6216\u4e0d\u6ee1\n        if cart_abandonment_rate &gt; 0.7:\n            churn_prob += 0.15\n            \n        # \u5ba2\u670d\u8054\u7cfb\u591a\u53ef\u80fd\u8868\u793a\u6709\u95ee\u9898\n        if customer_service_contacts &gt; 5:\n            churn_prob += 0.1\n            \n        # \u975e\u6d3b\u8dc3\u7528\u6237 (\u6708\u8bbf\u95ee\u6b21\u6570\u5c11)\n        if monthly_visits &lt; 3:\n            churn_prob += 0.1\n            \n        # Bronze\u4f1a\u5458\u672c\u8eab\u6d41\u5931\u7387\u53ef\u80fd\u7a0d\u9ad8\n        if membership_tier == 'Bronze':\n            churn_prob += 0.05\n            \n        # \u52a0\u5165\u968f\u673a\u6027\n        churn_prob += np.random.normal(0, 0.1)\n        churn_prob = np.clip(churn_prob, 0, 1) # \u9650\u5236\u57280-1\u4e4b\u95f4\n        \n        is_churned = int(np.random.random() &lt; churn_prob)\n        \n        data.append({\n            'user_id': user_id,\n            'age': age,\n            'gender': gender,\n            'membership_tier': membership_tier,\n            'location_city_tier': location_city_tier,\n            'signup_date': signup_date,\n            'total_orders': total_orders,\n            'total_spent': round(total_spent, 2),\n            'avg_order_value': round(avg_order_value, 2),\n            'days_since_last_purchase': round(days_since_last_purchase, 2),\n            'monthly_visits': monthly_visits,\n            'cart_abandonment_rate': round(cart_abandonment_rate, 4),\n            'customer_service_contacts': customer_service_contacts,\n            'discount_usage_freq': round(discount_usage_freq, 4),\n            'is_churned': is_churned\n        })\n        \n    df = pd.DataFrame(data)\n    # \u8ba1\u7b97\u884d\u751f\u7279\u5f81\n    df&#91;'signup_date'] = pd.to_datetime(df&#91;'signup_date'])\n    df&#91;'tenure_days'] = (datetime.now() - df&#91;'signup_date']).dt.days\n    df&#91;'tenure_days'] = df&#91;'tenure_days'].astype(int)\n    \n    csv_filename = f'{REPORT_PREFIX}_\u6a21\u62df\u6570\u636e.csv'\n    df.to_csv(csv_filename, index=False, encoding='utf-8-sig')\n    print(f\"\u6a21\u62df\u6570\u636e\u5df2\u751f\u6210\u5e76\u4fdd\u5b58\u81f3: {csv_filename}\")\n    return df\n\n# --- \u6570\u636e\u9884\u5904\u7406 ---\n\ndef preprocess_data(df):\n    \"\"\"\u6570\u636e\u9884\u5904\u7406\"\"\"\n    print(\"\\n--- \u6b63\u5728\u8fdb\u884c\u6570\u636e\u9884\u5904\u7406 ---\")\n    df_processed = df.copy()\n    \n    # 1. \u5904\u7406\u65e5\u671f\u7279\u5f81 (\u8fd9\u91cc\u5df2\u8f6c\u6362\u4e3a\u5929\u6570\uff0c\u65e0\u9700\u8fdb\u4e00\u6b65\u5904\u7406)\n    \n    # 2. \u7f16\u7801\u5206\u7c7b\u53d8\u91cf\n    le_gender = LabelEncoder()\n    le_membership = LabelEncoder()\n    le_location = LabelEncoder()\n    \n    df_processed&#91;'gender_encoded'] = le_gender.fit_transform(df_processed&#91;'gender'])\n    df_processed&#91;'membership_tier_encoded'] = le_membership.fit_transform(df_processed&#91;'membership_tier'])\n    df_processed&#91;'location_city_tier_encoded'] = le_location.fit_transform(df_processed&#91;'location_city_tier'])\n    \n    # 3. \u9009\u62e9\u7528\u4e8e\u5efa\u6a21\u7684\u7279\u5f81\u5217\n    feature_columns = &#91;\n        'age', 'gender_encoded', 'membership_tier_encoded', 'location_city_tier_encoded',\n        'total_orders', 'total_spent', 'avg_order_value', 'days_since_last_purchase',\n        'monthly_visits', 'cart_abandonment_rate', 'customer_service_contacts',\n        'discount_usage_freq', 'tenure_days'\n    ]\n    \n    X = df_processed&#91;feature_columns]\n    y = df_processed&#91;'is_churned']\n    \n    print(f\"\u9884\u5904\u7406\u5b8c\u6210\u3002\u7279\u5f81\u77e9\u9635\u5f62\u72b6: {X.shape}, \u6807\u7b7e\u5411\u91cf\u5f62\u72b6: {y.shape}\")\n    return X, y, le_gender, le_membership, le_location, feature_columns\n\n# --- \u6a21\u578b\u8bad\u7ec3\u4e0e\u8bc4\u4f30 ---\n\ndef train_and_evaluate_models(X, y):\n    \"\"\"\u8bad\u7ec3\u548c\u8bc4\u4f30\u591a\u4e2a\u6a21\u578b\"\"\"\n    print(\"\\n--- \u6b63\u5728\u8bad\u7ec3\u548c\u8bc4\u4f30\u6a21\u578b ---\")\n    \n    # \u5212\u5206\u8bad\u7ec3\u96c6\u548c\u6d4b\u8bd5\u96c6\n    X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=RANDOM_SEED, stratify=y)\n    \n    # \u7279\u5f81\u7f29\u653e (\u5bf9\u903b\u8f91\u56de\u5f52\u6bd4\u8f83\u91cd\u8981)\n    scaler = StandardScaler()\n    X_train_scaled = scaler.fit_transform(X_train)\n    X_test_scaled = scaler.transform(X_test)\n    \n    results = {}\n    \n    # --- 1. \u903b\u8f91\u56de\u5f52 ---\n    print(\"\u8bad\u7ec3\u903b\u8f91\u56de\u5f52\u6a21\u578b...\")\n    model_lr = LogisticRegression(random_state=RANDOM_SEED, max_iter=1000)\n    model_lr.fit(X_train_scaled, y_train)\n    y_pred_lr = model_lr.predict(X_test_scaled)\n    y_pred_proba_lr = model_lr.predict_proba(X_test_scaled)&#91;:, 1]\n    \n    results&#91;'Logistic Regression'] = {\n        'model': model_lr,\n        'predictions': y_pred_lr,\n        'probabilities': y_pred_proba_lr,\n        'accuracy': accuracy_score(y_test, y_pred_lr),\n        'precision': precision_score(y_test, y_pred_lr),\n        'recall': recall_score(y_test, y_pred_lr),\n        'f1': f1_score(y_test, y_pred_lr),\n        'auc': roc_auc_score(y_test, y_pred_proba_lr)\n    }\n    \n    # --- 2. \u968f\u673a\u68ee\u6797 ---\n    print(\"\u8bad\u7ec3\u968f\u673a\u68ee\u6797\u6a21\u578b...\")\n    model_rf = RandomForestClassifier(n_estimators=100, random_state=RANDOM_SEED, n_jobs=-1)\n    model_rf.fit(X_train, y_train) # \u968f\u673a\u68ee\u6797\u4e0d\u9700\u8981\u7279\u5f81\u7f29\u653e\n    y_pred_rf = model_rf.predict(X_test)\n    y_pred_proba_rf = model_rf.predict_proba(X_test)&#91;:, 1]\n    \n    results&#91;'Random Forest'] = {\n        'model': model_rf,\n        'predictions': y_pred_rf,\n        'probabilities': y_pred_proba_rf,\n        'accuracy': accuracy_score(y_test, y_pred_rf),\n        'precision': precision_score(y_test, y_pred_rf),\n        'recall': recall_score(y_test, y_pred_rf),\n        'f1': f1_score(y_test, y_pred_rf),\n        'auc': roc_auc_score(y_test, y_pred_proba_rf)\n    }\n    \n    # --- \u6bd4\u8f83\u548c\u62a5\u544a ---\n    print(\"\\n--- \u6a21\u578b\u6027\u80fd\u5bf9\u6bd4 ---\")\n    comparison_df = pd.DataFrame({\n        'Model': list(results.keys()),\n        'Accuracy': &#91;results&#91;k]&#91;'accuracy'] for k in results.keys()],\n        'Precision': &#91;results&#91;k]&#91;'precision'] for k in results.keys()],\n        'Recall': &#91;results&#91;k]&#91;'recall'] for k in results.keys()],\n        'F1-Score': &#91;results&#91;k]&#91;'f1'] for k in results.keys()],\n        'AUC-ROC': &#91;results&#91;k]&#91;'auc'] for k in results.keys()]\n    })\n    print(comparison_df.round(4).to_string(index=False))\n    \n    # \u9009\u62e9\u6700\u4f73\u6a21\u578b (\u4ee5AUC\u4e3a\u51c6)\n    best_model_name = comparison_df.loc&#91;comparison_df&#91;'AUC-ROC'].idxmax(), 'Model']\n    best_model = results&#91;best_model_name]&#91;'model']\n    best_predictions = results&#91;best_model_name]&#91;'predictions']\n    best_probabilities = results&#91;best_model_name]&#91;'probabilities']\n    \n    print(f\"\\n\u9009\u62e9\u6700\u4f73\u6a21\u578b: {best_model_name}\")\n    \n    # \u7ed8\u5236\u6700\u4f73\u6a21\u578b\u7684\u6df7\u6dc6\u77e9\u9635\n    cm = confusion_matrix(y_test, best_predictions)\n    plt.figure(figsize=(8, 6))\n    sns.heatmap(cm, annot=True, fmt='d', cmap='Blues', \n                xticklabels=&#91;'Not Churned', 'Churned'], \n                yticklabels=&#91;'Not Churned', 'Churned'])\n    plt.title(f'\u6df7\u6dc6\u77e9\u9635 - {best_model_name}')\n    plt.xlabel('\u9884\u6d4b\u6807\u7b7e')\n    plt.ylabel('\u771f\u5b9e\u6807\u7b7e')\n    cm_path = f'{REPORT_PREFIX}_\u6df7\u6dc6\u77e9\u9635_{best_model_name.replace(\" \", \"_\")}.png'\n    plt.savefig(cm_path)\n    plt.close()\n    print(f\"\u6df7\u6dc6\u77e9\u9635\u56fe\u8868\u5df2\u4fdd\u5b58\u81f3: {cm_path}\")\n    \n    # \u6253\u5370\u6700\u4f73\u6a21\u578b\u7684\u8be6\u7ec6\u5206\u7c7b\u62a5\u544a\n    print(f\"\\n--- {best_model_name} \u8be6\u7ec6\u5206\u7c7b\u62a5\u544a ---\")\n    print(classification_report(y_test, best_predictions, target_names=&#91;'Not Churned', 'Churned']))\n    \n    return results, best_model_name, scaler, X_test, y_test, cm_path\n\n# --- \u7279\u5f81\u91cd\u8981\u6027\u5206\u6790 ---\n\ndef analyze_feature_importance(model, feature_names, model_name):\n    \"\"\"\u5206\u6790\u5e76\u53ef\u89c6\u5316\u7279\u5f81\u91cd\u8981\u6027\"\"\"\n    print(f\"\\n--- \u5206\u6790 {model_name} \u7279\u5f81\u91cd\u8981\u6027 ---\")\n    \n    if hasattr(model, 'feature_importances_'):\n        # \u968f\u673a\u68ee\u6797\u7b49\u6811\u6a21\u578b\n        importances = model.feature_importances_\n        indices = np.argsort(importances)&#91;::-1]\n        title = f'{model_name} - \u7279\u5f81\u91cd\u8981\u6027 (\u57fa\u4e8e\u4e0d\u7eaf\u5ea6)'\n    elif hasattr(model, 'coef_'):\n        # \u7ebf\u6027\u6a21\u578b (\u903b\u8f91\u56de\u5f52\u7cfb\u6570\u7684\u7edd\u5bf9\u503c)\n        importances = np.abs(model.coef_&#91;0])\n        indices = np.argsort(importances)&#91;::-1]\n        title = f'{model_name} - \u7279\u5f81\u91cd\u8981\u6027 (\u57fa\u4e8e\u7cfb\u6570\u7edd\u5bf9\u503c)'\n    else:\n        print(\"\u6a21\u578b\u4e0d\u652f\u6301\u7279\u5f81\u91cd\u8981\u6027\u5206\u6790\u3002\")\n        return None, None\n\n    # \u521b\u5efa\u7279\u5f81\u91cd\u8981\u6027DataFrame\n    feature_importance_df = pd.DataFrame({\n        'feature': &#91;feature_names&#91;i] for i in indices],\n        'importance': &#91;importances&#91;i] for i in indices]\n    })\n    \n    print(\"\u7279\u5f81\u91cd\u8981\u6027\u6392\u5e8f:\")\n    print(feature_importance_df.head(10).to_string(index=False))\n\n    # \u7ed8\u5236\u7279\u5f81\u91cd\u8981\u6027\n    plt.figure(figsize=(10, 6))\n    sns.barplot(data=feature_importance_df.head(10), x='importance', y='feature', palette='viridis')\n    plt.title(title)\n    plt.xlabel('\u91cd\u8981\u6027')\n    plt.tight_layout()\n    feat_imp_path = f'{REPORT_PREFIX}_\u7279\u5f81\u91cd\u8981\u6027_{model_name.replace(\" \", \"_\")}.png'\n    plt.savefig(feat_imp_path)\n    plt.close()\n    print(f\"\u7279\u5f81\u91cd\u8981\u6027\u56fe\u8868\u5df2\u4fdd\u5b58\u81f3: {feat_imp_path}\")\n    \n    return feature_importance_df, feat_imp_path\n\n# --- \u62a5\u544a\u751f\u6210 ---\n\ndef generate_churn_prediction_report(best_model_name, results, cm_path, feat_imp_df, feat_imp_path):\n    \"\"\"\u751f\u6210\u6700\u7ec8\u7684\u6d41\u5931\u9884\u6d4b\u5206\u6790\u62a5\u544a\"\"\"\n    print(\"\\n--- \u6b63\u5728\u751f\u6210\u6d41\u5931\u9884\u6d4b\u5206\u6790\u62a5\u544a ---\")\n    from datetime import datetime\n    report_filename = f\"{REPORT_PREFIX}_{datetime.now().strftime('%Y%m%d_%H%M%S')}.txt\"\n    \n    best_metrics = results&#91;best_model_name]\n    \n    with open(report_filename, 'w', encoding='utf-8') as f:\n        f.write(\"=\" * 50 + \"\\n\")\n        f.write(\"        \u7535\u5546\u5e73\u53f0\u7528\u6237\u6d41\u5931\u9884\u6d4b\u5206\u6790\u62a5\u544a\\n\")\n        f.write(f\"        \u751f\u6210\u65f6\u95f4: {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}\\n\")\n        f.write(\"=\" * 50 + \"\\n\\n\")\n\n        f.write(\"--- 1. \u9879\u76ee\u6982\u8ff0 ---\\n\")\n        f.write(\"\u672c\u9879\u76ee\u65e8\u5728\u901a\u8fc7\u5206\u6790\u7528\u6237\u5386\u53f2\u884c\u4e3a\u6570\u636e\uff0c\u6784\u5efa\u673a\u5668\u5b66\u4e60\u6a21\u578b\u6765\u9884\u6d4b\u7528\u6237\u6d41\u5931\u7684\u53ef\u80fd\u6027\u3002\\n\")\n        f.write(\"\u76ee\u6807\u662f\u8bc6\u522b\u51fa\u9ad8\u98ce\u9669\u7528\u6237\uff0c\u4ee5\u4fbf\u8fd0\u8425\u56e2\u961f\u53ef\u4ee5\u63d0\u524d\u91c7\u53d6\u5e72\u9884\u63aa\u65bd\uff0c\u964d\u4f4e\u7528\u6237\u6d41\u5931\u7387\u3002\\n\\n\")\n\n        f.write(\"--- 2. \u6570\u636e\u6982\u89c8 ---\\n\")\n        f.write(\"\u6570\u636e\u6765\u6e90: \u6a21\u62df\u751f\u6210\u7684\u7535\u5546\u5e73\u53f0\u7528\u6237\u6570\u636e\u3002\\n\")\n        f.write(\"\u5173\u952e\u5b57\u6bb5: \u7528\u6237ID, \u5e74\u9f84, \u6027\u522b, \u4f1a\u5458\u7b49\u7ea7, \u57ce\u5e02\u7ea7\u522b, \u603b\u8ba2\u5355\u6570, \u603b\u6d88\u8d39\u989d, \u5e73\u5747\u8ba2\u5355\u4ef7\u503c,\\n\")\n        f.write(\"          \u8ddd\u79bb\u4e0a\u6b21\u8d2d\u4e70\u5929\u6570, \u6708\u5747\u8bbf\u95ee\u6b21\u6570, \u8d2d\u7269\u8f66\u653e\u5f03\u7387, \u5ba2\u670d\u8054\u7cfb\u6b21\u6570, \u6298\u6263\u4f7f\u7528\u9891\u7387, \u6ce8\u518c\u65f6\u957f, \u6d41\u5931\u6807\u7b7e\u3002\\n\")\n        f.write(\"\u539f\u59cb\u6570\u636e\u5df2\u4fdd\u5b58\u4e3a CSV \u6587\u4ef6\u3002\\n\\n\")\n\n        f.write(\"--- 3. \u6a21\u578b\u6027\u80fd\u8bc4\u4f30 ---\\n\")\n        f.write(\"\u8bad\u7ec3\u4e86\u4e24\u79cd\u6a21\u578b\uff1a\u903b\u8f91\u56de\u5f52 (Logistic Regression) \u548c \u968f\u673a\u68ee\u6797 (Random Forest)\u3002\\n\")\n        f.write(\"\u8bc4\u4f30\u6307\u6807\u5305\u62ec: \u51c6\u786e\u7387 (Accuracy), \u7cbe\u786e\u7387 (Precision), \u53ec\u56de\u7387 (Recall), F1\u5206\u6570 (F1-Score), AUC-ROC\u3002\\n\\n\")\n        \n        f.write(\"\u5404\u6a21\u578b\u6027\u80fd\u5bf9\u6bd4:\\n\")\n        comparison_df = pd.DataFrame({\n            'Model': list(results.keys()),\n            'Accuracy': &#91;results&#91;k]&#91;'accuracy'] for k in results.keys()],\n            'Precision': &#91;results&#91;k]&#91;'precision'] for k in results.keys()],\n            'Recall': &#91;results&#91;k]&#91;'recall'] for k in results.keys()],\n            'F1-Score': &#91;results&#91;k]&#91;'f1'] for k in results.keys()],\n            'AUC-ROC': &#91;results&#91;k]&#91;'auc'] for k in results.keys()]\n        })\n        f.write(comparison_df.round(4).to_string(index=False))\n        f.write(\"\\n\\n\")\n        \n        f.write(f\"\u6700\u4f73\u6a21\u578b: {best_model_name}\\n\")\n        f.write(f\"\u8be5\u6a21\u578b\u5728\u6d4b\u8bd5\u96c6\u4e0a\u7684\u8868\u73b0:\\n\")\n        f.write(f\"  - \u51c6\u786e\u7387 (Accuracy): {best_metrics&#91;'accuracy']:.4f}\\n\")\n        f.write(f\"  - \u7cbe\u786e\u7387 (Precision): {best_metrics&#91;'precision']:.4f}\\n\")\n        f.write(f\"  - \u53ec\u56de\u7387 (Recall): {best_metrics&#91;'recall']:.4f}\\n\")\n        f.write(f\"  - F1\u5206\u6570 (F1-Score): {best_metrics&#91;'f1']:.4f}\\n\")\n        f.write(f\"  - AUC-ROC: {best_metrics&#91;'auc']:.4f}\\n\")\n        f.write(\"\u6df7\u6dc6\u77e9\u9635\u8be6\u7ec6\u5206\u6790\u4e86\u6a21\u578b\u7684\u9884\u6d4b\u7ed3\u679c\uff0c\u56fe\u8868\u5df2\u751f\u6210\u3002\\n\")\n        f.write(f\"\u6df7\u6dc6\u77e9\u9635\u56fe\u8868: {cm_path}\\n\\n\")\n\n        f.write(\"--- 4. \u5173\u952e\u9a71\u52a8\u56e0\u7d20\u5206\u6790 ---\\n\")\n        f.write(\"\u901a\u8fc7\u5206\u6790\u6700\u4f73\u6a21\u578b\u7684\u7279\u5f81\u91cd\u8981\u6027\uff0c\u8bc6\u522b\u51fa\u5f71\u54cd\u7528\u6237\u6d41\u5931\u7684\u6700\u5173\u952e\u56e0\u7d20\u3002\\n\")\n        f.write(\"\u7279\u5f81\u91cd\u8981\u6027\u6392\u5e8f (Top 10):\\n\")\n        f.write(feat_imp_df.head(10).to_string(index=False))\n        f.write(\"\\n\u4ece\u4e0a\u8868\u53ef\u4ee5\u770b\u51fa\uff0c\u54ea\u4e9b\u7528\u6237\u884c\u4e3a\u548c\u5c5e\u6027\u5bf9\u6d41\u5931\u5f71\u54cd\u6700\u5927\u3002\\n\")\n        f.write(f\"\u7279\u5f81\u91cd\u8981\u6027\u56fe\u8868: {feat_imp_path}\\n\\n\")\n\n        f.write(\"--- 5. \u4e1a\u52a1\u5e94\u7528\u4e0e\u5efa\u8bae ---\\n\")\n        f.write(\"1. \u9ad8\u98ce\u9669\u7528\u6237\u8bc6\u522b: \u5229\u7528\u6700\u4f73\u6a21\u578b\u5bf9\u6240\u6709\u7528\u6237\u8ba1\u7b97\u6d41\u5931\u6982\u7387\uff0c\u7b5b\u9009\u51fa\u9ad8\u98ce\u9669\u7528\u6237\u5217\u8868\u3002\\n\")\n        f.write(\"2. \u7cbe\u51c6\u5e72\u9884:\\n\")\n        f.write(\"   - \u5bf9\u4e8e'\u8ddd\u79bb\u4e0a\u6b21\u8d2d\u4e70\u5929\u6570'\u957f\u7684\u7528\u6237\uff0c\u53ef\u63a8\u9001\u53ec\u56de\u4f18\u60e0\u5238\u3002\\n\")\n        f.write(\"   - \u5bf9\u4e8e'\u8d2d\u7269\u8f66\u653e\u5f03\u7387'\u9ad8\u7684\u7528\u6237\uff0c\u53ef\u5206\u6790\u652f\u4ed8\u6d41\u7a0b\u6216\u63d0\u4f9b\u5ba2\u670d\u5e2e\u52a9\u3002\\n\")\n        f.write(\"   - \u5bf9\u4e8e'\u6708\u5747\u8bbf\u95ee\u6b21\u6570'\u5c11\u7684\u7528\u6237\uff0c\u53ef\u901a\u8fc7\u90ae\u4ef6\/SMS\u63a8\u9001\u4e2a\u6027\u5316\u5185\u5bb9\u3002\\n\")\n        f.write(\"3. \u4ea7\u54c1\u4e0e\u8fd0\u8425\u4f18\u5316:\\n\")\n        f.write(\"   - \u6839\u636e\u5173\u952e\u7279\u5f81\u4f18\u5316\u7528\u6237\u5f15\u5bfc\u548c\u7559\u5b58\u7b56\u7565\u3002\\n\")\n        f.write(\"   - \u9488\u5bf9\u4e0d\u540c\u4f1a\u5458\u7b49\u7ea7\u5236\u5b9a\u5dee\u5f02\u5316\u7684\u5fe0\u8bda\u5ea6\u8ba1\u5212\u3002\\n\")\n        f.write(\"4. \u6a21\u578b\u8fed\u4ee3: \u5b9a\u671f\u4f7f\u7528\u6700\u65b0\u6570\u636e\u91cd\u65b0\u8bad\u7ec3\u6a21\u578b\uff0c\u4ee5\u9002\u5e94\u5e02\u573a\u548c\u7528\u6237\u884c\u4e3a\u7684\u53d8\u5316\u3002\\n\\n\")\n\n        f.write(\"=\" * 50 + \"\\n\")\n        f.write(\"                    \u62a5\u544a\u7ed3\u675f\\n\")\n        f.write(\"=\" * 50 + \"\\n\")\n\n    print(f\"\u6d41\u5931\u9884\u6d4b\u5206\u6790\u62a5\u544a\u5df2\u751f\u6210: {report_filename}\")\n\n# --- \u4e3b\u51fd\u6570 ---\n\ndef main():\n    \"\"\"\u4e3b\u51fd\u6570\"\"\"\n    # 1. \u751f\u6210\u6570\u636e\n    df_churn = generate_sample_churn_data(NUM_USERS)\n    \n    # 2. \u6570\u636e\u9884\u5904\u7406\n    X, y, le_gender, le_membership, le_location, feature_cols = preprocess_data(df_churn)\n    \n    # 3. \u6a21\u578b\u8bad\u7ec3\u4e0e\u8bc4\u4f30\n    results, best_model_name, scaler, X_test, y_test, cm_path = train_and_evaluate_models(X, y)\n    \n    # 4. \u7279\u5f81\u91cd\u8981\u6027\u5206\u6790 (\u9488\u5bf9\u6700\u4f73\u6a21\u578b)\n    best_model = results&#91;best_model_name]&#91;'model']\n    feat_imp_df, feat_imp_path = analyze_feature_importance(best_model, feature_cols, best_model_name)\n    \n    # 5. \u751f\u6210\u62a5\u544a\n    generate_churn_prediction_report(best_model_name, results, cm_path, feat_imp_df, feat_imp_path)\n    \n    print(\"\\n\u7528\u6237\u6d41\u5931\u9884\u6d4b\u5206\u6790\u6d41\u7a0b\u5b8c\u6210\u3002\")\n\nif __name__ == \"__main__\":\n    main()\n<\/code><\/pre>\n","protected":false},"excerpt":{"rendered":"<p>\u5f00\u53d1\u601d\u8def:\u6a21\u62df\u4e865000\u540d\u7528\u6237\u7684\u6570\u636e\uff0c\u5305\u542b\u7528\u6237\u753b\u50cf\uff08\u5e74\u9f84\u3001\u6027\u522b\u3001\u4f1a\u5458\u7b49\u7ea7\u3001\u57ce\u5e02\u7ea7\u522b\uff09\u3001\u884c\u4e3a\u7279\u5f81\uff08\u8ba2\u5355&hellip; <a href=\"http:\/\/viplao.com\/index.php\/2025\/09\/13\/%e3%80%90python%e5%ae%9e%e8%b7%b5%e6%a1%88%e4%be%8b%e3%80%91%e7%94%b5%e5%95%86%e5%b9%b3%e5%8f%b0%e6%95%b0%e6%8d%ae%e5%88%86%e6%9e%90%e5%92%8c%e6%8c%96%e6%8e%98-%e7%94%a8%e6%88%b7%e6%b5%81%e5%a4%b1\/\" class=\"more-link read-more\" rel=\"bookmark\">\u7ee7\u7eed\u9605\u8bfb <span class=\"screen-reader-text\">\u3010Python\u5b9e\u8df5\u6848\u4f8b\u3011\u7535\u5546\u5e73\u53f0\u6570\u636e\u5206\u6790\u548c\u6316\u6398 &#8211; \u7528\u6237\u6d41\u5931\u9884\u6d4b<\/span><i class=\"fa fa-arrow-right\"><\/i><\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":[],"categories":[1],"tags":[28],"views":345,"_links":{"self":[{"href":"http:\/\/viplao.com\/index.php\/wp-json\/wp\/v2\/posts\/3895"}],"collection":[{"href":"http:\/\/viplao.com\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/viplao.com\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/viplao.com\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/viplao.com\/index.php\/wp-json\/wp\/v2\/comments?post=3895"}],"version-history":[{"count":2,"href":"http:\/\/viplao.com\/index.php\/wp-json\/wp\/v2\/posts\/3895\/revisions"}],"predecessor-version":[{"id":3912,"href":"http:\/\/viplao.com\/index.php\/wp-json\/wp\/v2\/posts\/3895\/revisions\/3912"}],"wp:attachment":[{"href":"http:\/\/viplao.com\/index.php\/wp-json\/wp\/v2\/media?parent=3895"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/viplao.com\/index.php\/wp-json\/wp\/v2\/categories?post=3895"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/viplao.com\/index.php\/wp-json\/wp\/v2\/tags?post=3895"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}