Data Science, Python, R
Resume:
ZHEYI XU
EDUCATION
COURSEWORK & PROJECTS
PROFESSIONAL EXPERIENCE
SKILLS & COMPETENCIES
UNIVERSITY OF MICHIGAN - ANN ARBOR
Master of Science in Applied Statistics GPA: 3.95/4.00 Ann Arbor, MI
Dec. 2015
• Memberships: Michigan Student Artificial Intelligence Lab, Michigan Data Science Team, American Statistical Association. ZHEJIANG UNIVERSITY
Bachelor of Science in Mathematics & Applied Mathematics (Honors) GPA: 3.70/4.00 Hangzhou, China
Jun. 2014
• Honors & Awards: Academic Elite Scholarship (Top 3%); Basic Sciences Scholarship in Mathematics (Top 3%).
• Memberships: Chu-Kochen Honors College, National Training Program in Basic Sciences.
• Exchange Program: Dept. Math, Pennsylvania State University in Fall 2013, Bachelor honors courses (GPA: 4.00/4.00).
• Statistics: Linear Models (Regression), Analysis of Multivariate and Categorical Data (Machine Learning), Statistical Inference, Data Manipulation (Python), Applied Stochastic Modeling, Probability Theory, Large Sample Theory, Statistics for Financial Data, Econometrics, Statistical Consulting.
• Mathematics: Multivariate Calculus, Advanced Algebra, Real Analysis, Complex Analysis, Functional Analysis, Combinatorial Analysis, Differential Equations, Differential Manifolds, Financial Engineering (Arbitrage Theory).
• Projects:
- Assessment of Surgical Quality: Sep. 2015 - Dec. 2015
- Evaluated the quality of thyroid lobectomy and thyroidectomy over 52 hospitals participated in Michigan Surgical Quality Collaborative by several methods: stratified model, counterfactual model, matching and historical comparison.
- Prediction of Bank Telemarketing: Jan. 2015 - Apr. 2015
- Built models in R to predict successful bank deposit subscription from direct marketing campaign data of a Portuguese bank, by different machine learning techniques: logistic regression, naive bayesian classifier, SVM, linear discriminant analysis, decision tree and neutral network, and evaluated their accuracy.
- Algorithms Research & Parallel Computing: Sep. 2014 - Dec. 2014
- Applied sequential stochastic gradient descent algorithm into the optimization procedure of kernel SVM in C++; Enabled a distributed storage on 8, 16, 20 processors using MPI to reduce running time; Analyzed time complexity and compared accuracy on datasets of different sizes with widely-used R packages: ksvm, SVM-light. COOK COUNTY SHERIFF’S OFFICE
Research Fellow, Department of Research
Chicago, IL
Jun. 2015 - Aug. 2015
• Statistical Modeling: Constructed negative binomial model and regression tree model with passage of time to predict individual’s length of stay in Cook County Jail from past 5 years data; Cleaned and selected predictive features using Stata.
• Time Series Analysis: Predicted 7-week jail population from historical weekly data within 15% error by building models with ARMA (autoregressive moving average) error and seasonal component on both jail admissions and releases.
• Data Visualization: Created charts and graphics presenting descriptive statistics and analytic results using ggplot2.
• Text Parsing: Extracted sentence lengths and reasons using Python from unconstructed data in criminal database.
• Policy Recommendations: Presented suggestions on file digitization, database management and update, court schedule optimization and advanced policy research methodologies. KUANG-CHI INSTITUTE OF ADVANCED TECHNOLOGY (Telecom Research Institute) Research Engineer Intern, R&D Center
Shenzhen, China
Jul. 2013 - Aug. 2013
• Algorithms Study: Tested efficiency of 6 basic optimizers; Examined multi-objective algorithms: MOGT (Multi- Objective Optimization based on Game Theory), MOGA-II (Multi-Objective Genetic Algorithm with Elitism).
• Multi-Objective Optimization: Optimized 3 objects simultaneously: far-field realized gain, main lobe degree and side lobe level of a 5-by-5 phased array antenna at fix frequency by adjusting geometric parameters; Minimized the error rate in degree and side lobe level by 20.5% and 7.5%, maximized far-field realized gain by 37.6%.
• Initiative: Established the interface between CST (electromagnetic simulation software) and modeFrontier (optimization software) to conduct multi-objective optimization; Wrote an instruction manual on parameter mapping.
• Programming Skills: R, Python, C/C++, Stata, SQL, SAS, APIs, Spark, MPI, Unix Shell, LaTeX, Matlab. 2240 St.Francis Dr., Apt. 101
Ann Arbor, MI 48104
******@*****.***
Contact this candidate