Luyu Li

Luyu Li

Followers of Luyu Li2000 followers
location of Luyu LiHaidian District, Beijing, China

Connect with Luyu Li to Send Message

Connect

Connect with Luyu Li to Send Message

Connect
  • Timeline

  • About me

    PHD candidate at Tsinghua University

  • Education

    • Columbia University

      2019 - 2021
      Master of Science Applied Analytics
    • Peking university

      2017 - 2019
      Bachelor economics GPA 3.5/4.0
    • 清华大学经济管理学院 Tsinghua University School of Economics and Management

      2023 - 2027
      博士 营销
    • 国际关系学院

      2014 - 2018
      学士 Communication, General
    • Tsinghua

      2023 - 2027
      博士 Marketing/Marketing Management, General
  • Experience

    • 北京大学

      Jan 2018 - Jan 2019

      使用机器学习的方法解决自然语言处理的问题,采用朴素贝叶斯的方法解决分类问题。编写学习模型,并把模型在原数据上的运行结果和人工结果做对比。后期把模型运用在所有数据上,自动得出没有人工结果的数据的答案。计算流程如下:1. 解析输入输出2. 中文分词3. 用 TF-IDF 模型提取特征4. 用朴素贝叶斯模型做机器学习5. 用学习完毕的模型做推断项目属于北京大学企业大数据研究中心ESIEC项目组,共有在甘肃、河南、辽宁、上海、广东进行田野调查收集的 58500份问卷,目的是形成中国中小企业生存状况数据库。由于问卷第一个选择题——企业联系状况的两个选项(不能确定、找不到)包含了非常复杂的情况(企业虚拟注册、个别访员未努力尝试、企业可能存在但拒访、企业未在注册地址经营、企业搬迁等),必须通过访员写的文本备注确定企业存续状况,因此建立了这个模型进行自然语言处理。随机抽取了其中的 2000 份,人工阅读问题输入,并选择问题输出,作为训练数据。使用一些中文分词的程序包,把输入文本变成Bag of word 数据,然后作为朴素贝叶斯机器学习包的输入。训练好模型后,应用在所有数据上,实现58500份问卷的输出。 Show less

      • 数据清洗助研

        Aug 2018 - Jan 2019
      • 访员

        Jun 2018 - Dec 2018
      • 助教

        Jan 2018 - Jun 2018
      • intern

        Nov 2017 - Jun 2018
      • Research Assistant

        Jan 2018 - Apr 2018
    • EverChain

      Aug 2018 - Mar 2019
      AI & data analysis internship

      ·为想要约会的用户提供好友匹配和自动推送服务:使用AI算法进行计算,根据结果为用户提供合适的对象并自动推送,实现用户匹配·使用python进行用户数据分析、小程序访问来源分析·熟悉elk,使用kibana 进行es展示·使用fineBI可视化分析mysql导入的文件, 实现实时更新可视化结果

    • Chinese Academy of Social Sciences

      Feb 2019 - Sept 2019
      Research Assistant

      Data cleaning;Analyzed the insufficiently transparent raw data of the 2018 refined EBA model by IMF based on the regression results and the index value;Deal with the newly adjusted EBA model in the 2018 “External Imbalance Report” of the International Monetary Fund, and calculate the original data based on indicators, models and regression coefficients (lower reporting data transparency)

    • 北大

      Mar 2019 - Nov 2019
      研究助理

      北京大学教育学院-蒋承副教授-科研助理·高中教育/高考成绩影响因素定量研究

    • Independent project

      May 2020 - Mar 2021

      Residents of new houses in the community often complain to the community property of various problems, such as housing problems that require minor repairs, missing parcels sent, collision of public entertainment room booking time, late management fees of the community, and poor communication of property information. The community has a professional management company in charge of the community management, but the labor efficiency is low, the problem is still frequent and the business methods are very conservative. In response to this problem, the community hopes to develop a management system to provide a three-party collaborative communication platform for community properties, third-party management companies and community residents to help improve community management efficiency and solve property problems. Features:DashBoard – designed to announce upcoming events, alert, monthly newsletter and a condo policyDiscuss Board – designed for the resident to post anecdotes, concerns or whatever they want, for example, I found the light in lower garage is out of workChat Thread – designed for the trustee to communicate instead of using email thread, the idea is coming from slackCalendar Schedule – designed to book a common room, reserve maintain service for the condo, such as dumpster clean, garage clean, fire alarm testing, or elevator inspection, etc.Payment tool – designed to pay condo fee, common room reserve fee (but here we should concern about the data safety and privacy issue) Show less Designed and implemented an interactive web application for users to search and apply available job positions.Performed front-end web UI design and implementation using HTML/CSS/JavaScript. Implemented RESTful APIs using Java servlets, retrieved job descriptions using Github API and stored data in MySQL.Explored multiple recommendation algorithms and extracted keywords from job descriptions to implement a Content-based algorithm.Deployed the service to AWS EC2 and performed load tests using Apache JMeter. Show less

      • carried on a React project

        May 2020 - Mar 2021
      • Developed Community Property Management System

        Jul 2020 - Nov 2020
      • carried on a project: building a short video web app with Go, Gin and React

        Apr 2020 - Aug 2020
      • carried on a project: Job+: A Personalized Job Recommendation Engine

        Apr 2020 - Jul 2020
      • Developed Tinnews: a Tinder-like News App

        May 2020 - Jun 2020
  • Licenses & Certifications