Data State University

HAO YI

*** ***** **.

Edison, NJ *****

Tel: 732-***-****

Email: ********@*******.***

Education

Rutgers, THE STATE UNIVERSITY OF NEW JERSEY New Brunswick, NJ

Master of Science, Mathematical Finance, May 2013-2015

UNIVERSITY OF LIVERPOOL Liverpool, UK

Bachelor of Science, Mathematics with Finance, July 2013 (First Class, top 10%)

Experience

BANK OF CHINA SuZhou, China

Data Analyst Intern Summer, 2012

Collected, summarized and analyzed client data as well as financial statements

•

Responded to questions from clients; Prepared briefings for senior managers

•

Projects

Twitter Sentiment Analysis, Data Mining

-Investigate the predictive power of public sentiment from various Twitter accounts for stock market return movement

-Request twitter data through Twitter API

-Data cleaning and feature extraction to raw collected data in R

-Tweets sentiment classification (supervised machine learning) comparison among Multinomial Naïve Bayes, Support

Vector Machine and Logistic Regression using Scikit-learn and Pandas packages in Python

-Granger causality analysis of classified sentiment on AR (2) time series model

Movie Recommendation Database, Business Data Management

-Create ‘Movie Recommendation’ relational schema based on conceptual ER-diagram in MAMP MySQL on Toad

-Verify and normalize the database system to Third Normal Form (3NF)

-Apply SQL query for specific users’ requirements and output recommended movies

News grouping, Text Classification (Scikit-learn package)

-Train a Naïve Bayes classifier composed of a feature vectorizer (transform text into numeric features), evaluated by K-

fold cross-validation

-Improve the results by trying to sparse the text tokens and remove stop words

-Model selection for the best ‘alpha’ parameter in MultinomialNB and ‘gamma’ in SVC; ‘C-gamma’ combination of

parameter selection using Grid Search

Rating Cereals, Regression Analysis

-Build multiple linear regression model based on three basic assumptions via R software

-Determine linearity between transformed response and features (regressors) using partial regression plots

-Variable selection employed by Backward selection, AIC criterion and Mallow’s Cp Statistic

-Test for adequacy among candidate models, including basic assumptions and multicollinearity

-Conclude the best model and detect any potential drawbacks

Stock data set analysis, Time Series Analysis

-E-views associated with time series analysis, including stationary analysis, seasonality analysis, univariate modeling,

cointegration analysis, building error correction model, causality analysis, and K-S normality test

Additional

Working knowledge of MS Office Suite, Scikit-learn and Pandas packages in Ipython Notebook, R, E-views

•

Basic knowledge of SAP, Data Warehouse, Excel Pivot, MATLAB, C++, C#(ASP.NET&ADO.NET)

•

Contact this candidate