johnsmith0209 / p2pRisk

Breif

This REPO is for the homework of BIT's MEM course, and has no intention for business usage.

This is a project that collects peer to peer loaning platforms' data and analysis the risk of it.

Web Spider

• Tool: Node.js, Electron
• Folder: ./spider-wdzj
• source: 网贷之家(wdzj)
• Main API:
  1. Platforms' Basic Data: https://shuju.wdzj.com/plat-data-custom.html
  2. Platforms' Broken/Failure Data: https://shuju.wdzj.com/problem-list-all.html
• Run:
  1. cd ./spider-wdzj && npm start
  2. Login manually, cause it could be a complicated work if I implements a method do it programmatically, could do it in the future.
  3. Electron will open dev tool automatically,
  4. Then type getAllBasic(), this will fetch data of platforms which dates ranges from 2017-01-01 to now(). It will generate a JSON file, ./datas/basics.json.
  5. Then type getBroken(), this will fetch data of all the platforms which are/were malfunctioning. This fetching will ignore date ranges and return all dates' data.
  6. Finally, call attachBroken(), this function will attach the data from getBroken into the basics data. then get our finall data structure as below;

{
  "$wdzjPlatId": {
    "wdzjPlatId": int, // wdzj platform id: 40 
    "platName": string, // platform name: 人人贷
    "background": string,
    "newbackgroud": string, // 国资
    "onlineTime": Date, // YYYY-MM-DD
    "businessTypes": string[],
    "businessTypeIdS": int[],
    "source": "API",
    "basic": [
      {
        "amount": decimal(,2),
        "avgBidMoney": decimal(,2),
        "avgBorrowMoney": decimal(,2),
        "bidderNum": int,
        "borrowerNum": int,
        "currentLeverageAmount": decimal(,2),
        "dataStatus": int,
        "developZhishu": int,
        "differStatus": int,
        "endDate": Date, // YYYY-MM-DD
        "fullloanTime": decimal(,2),
        "incomeRate": decimal(,2),
        "loanPeriod": decimal(,2),
        "netInflowOfThirty": decimal(,2),
        "prizeTypeDetail": string,
        "regCapital": int,
        "startDate": Date, // YYYY-MM-DD
        "status": int,
        "stayStillOfNextSixty": decimal(,2),
        "stayStillOfTotal": decimal(,2),
        "timeOperation": int,
        "top10DueInProportion": decimal(,2),
        "top10StayStillProportion": decimal(,2),
        "totalLoanNum": int,
        "weightedAmount": int,
        "month": Date, // YYYY-MM-DD
      }
    ...],
    // if it borken, then this field will not be empty
    "broken": {
      "proTime": Date, // YYYY-MM
      "problemTime": Date, // YYYY-MM-DD
      "type": string, // 刑侦介入 broken type
      "regCapital": int, // registration capital amount
    }
  }
}

TODO

• Use python to implement Few-shot learnning to do the classification & risk evaluation.