Data Scientist

Yuan Liu

Seattle, Washington 614-***-**** no sponsorship required **************@*****.***

w www.linkedin.com/in/liuyuan1129 https://github.com/yuanyvette1129 SKILLS

Programming Languages: Python, SQL, SAS certified

Technologies: - Apache Spark, MapReduce, Hadoop, Databricks, Google Colab, Git, Tableau

- Machine Learning: Random Forest, Gradient Boosted Trees, Logistic Regression, Lasso, Ridge, PCA, KNN, NLP, Clustering, Anomaly Detection, Recommendation System

- Statistics: A/B Testing, Experimental Design, Econometric Models, Time Series Forecasting EXPERIENCE

Data Science Fellow, Insight, Seattle WA May 2020 - Present

● Consulted for a mobile game company (A Thinking Ape) to monitor shard health and improve game monetization

● Preprocessed over 1.2 million game behavior data and identified 16 key gameplay features from over 90 features using filter and embedded feature selection techniques

● Built a random forest model that predicts player’s LTV by day 60 and identified low-performing shards for the community specialist team for early intervention

Research Assistant, George Washington Institute of Public Policy, Washington D.C. Sep 2015 - May 2020

● Collected income data from the Census of Population and visualized to understand the geography of U.S. county-level income inequality from 1950 to 2010

● Empirically evaluated the employment impacts of Florida county recycling programs based on fixed effects regression model and estimated that a 1 percentage point increase in county recycling rate leads to a 0.4% job growth in solid waste and recycling industry

Research Analyst, Ohio Civil Service Employee Association, Westerville, Ohio 2014 – 2015

● Provided data support for successful negotiation of a $1.5 million contract with County Job and Family Services

● Prepared charts using Tableau about wintertime slips and falls within Department of Rehabilitation and Corrections to help analyze and develop recommendations for injury reduction PROJECTS

User Segmentation and Churn Prediction based on 1.6 million Merchant Transaction Activities Feb 2021

● Identified 4 types of businesses using RFM clustering, to facilitate customized action plans to different segments

● Defined a churned merchant as not make a transaction after 30 days of their first transaction, and self-labeled merchants into churn/no churn

● Built a Random Forest model to predict which active merchants are likely to churn in the near future, with a F1 score= 0.45; found that the total transaction amount in a month is the most predictive of churning Financial Anomaly Detection and Risk Analysis in Python August 2020

● Developed a machine learning model and built an alert system to predict and prevent fraudulent activities

● Performed exploratory data analysis on 138K+ transactions and preprocessed data by matching IP address to country, feature engineering, encoding categorical features, and handling imbalance labeled data by SMOTE

● Applied distribution-based modeling and supervised machine learning algorithms, and selected random forest model as the final model (best F1: 0.67)

● Found that more than 50% of fraud activities occurred 1 second after sign up and the more a device or an IP is shared, the more likely to be classified as at risk EDUCATION

George Washington University 2015 - Present

Ph.D. candidate (ABD), Public Policy and Public Administration Ohio State University 2012 - 2014

Master of Public Administration & Graduate Minor in Statistics Laioffer 2019 - 2020

Artificial Intelligence & Data Engineering Certificate

Contact this candidate