{
  "fetched_at": "2026-10-20T06:52:04Z",
  "source_url": "https://www.kaggle.com/competitions/fraud-detection-in-insurance-claims/data",
  "pages": [
    {
      "name": "data-description",
      "content": "<h2>文件描述</h2>\n<ul>\n<li><strong>train.csv</strong> - 训练集</li>\n<li><strong>test.csv</strong> - 测试集</li>\n<li><strong>sampleSubmission.csv</strong> - 提交样例</li>\n按照测试集的顺序进行提交，提交字段：policy_number, fraud_reported",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Description",
      "content": "<h1>问题描述</h1>\n<p>\n通过收集之前保单的信息，标记了保单是否为欺诈。现在保险公司想要请你创建一个保单预测模型，对于新的保单判断是否为欺诈保单。保单的字段包括了（客户信息、保单信息、车险的投保汽车信息、警察记录报告，出险情况等等）<br>\n请你通过模型，给出在测试集上的预测结果（1或者0）\n</p>\n<h1>数据说明</h1>\n训练数据和测试数据，保存在train.csv和test.csv文件中<br/>\n训练集包括700条记录，40个字段，字段说明如下：  <br/>\n| 字段 | 说明 |\n| --- | :--------: |\n|months_as_customer|成为客户的时长，以月为单位|\n|age|年龄|\n|policy_number|保险编号|\n|policy_bind_date|保险绑定日期|\n|policy_state|上保险所在地区|\n|policy_csl|组合单一限制Combined Single Limit |\n|policy_deductable|保险扣除额|\n|policy_annual_premium|每年的保费|\n|umbrella_limit|保险责任上限|\n|insured_zip|被保人邮编|\n|insured_sex|被保人姓名：FEMALE或者MALE|\n|insured_education_level|被保人学历|\n|insured_occupation|被保人职业|\n|insured_hobbies|被保人兴趣爱好|\n|insured_relationship|被保人关系|\n|capital-gains|资本收益|\n|capital-loss|资本损失|\n|incident_date|出险日期|\n|incident_type|出险类型|\n|collision_type|碰撞类型|\n|incident_severity|事故严重程度|\n|authorities_contacted|联系了当地的哪个机构|\n|incident_state|出事所在的州|\n|incident_city|出事所在的城市|\n|incident_location|出事所在的具体位置|\n|incident_hour_of_the_day|出事所在的小时（一天24小时的哪个时间）|\n|number_of_vehicles_involved|涉及的车辆数|\n|property_damage|是否有财产损失|\n|bodily_injuries|身体伤害|\n|witnesses|目击证人|\n|police_report_available|是否有警察记录的报告|\n|total_claim_amount|整体索赔金额|\n|injury_claim|伤害索赔金额|\n|property_claim|财产索赔金额|\n|vehicle_claim|汽车索赔金额|\n|auto_make|汽车品牌，比如Audi, BMW, Toyota, Volkswagen|\n|auto_model|汽车型号，比如A3,X5,Camry,Passat等|\n|auto_year|汽车购买的年份|\n|fraud_reported|是否欺诈，1或者0|\n|_c39]|无意义|\n\n\n测试数据主要包括300条记录，39个字段，不包括保险是否是欺诈的记录，需要通过由训练数据所建立的模型以及所给的测试数据，得出相应的保单是否是欺诈的预测。\n\n\n<p>评分标准：AUC</p>\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2445437%2Fcb3da0eb50c880f8a0649a34e525bf1d%2Fauc-1.png?generation=1575603193502330&alt=media)",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Evaluation",
      "content": "<p>评价指标：[AUC](http://en.wikipedia.org/wiki/Receiver_operating_characteristic)，即ROC曲线下面的面积</p>\n\n<h2>提交文件格式</h2>\n<p>提交文件包括了两列: policy_number 和 fraud_reported</p>\npolicy_number,fraud_reported<br>698589,1<br>287489,1<br>211578,0<br>...,...<br>914815,0<br>\n",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "rules",
      "content": " - Don't cheat!\n - Apply yourself!\n - Have fun!",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    }
  ]
}