SKILLS:

Proficient in: R, SAS, Python, SQL, STATA, JMP, VBA

Proficient in: Excel, Numbers, PowerPoint, Keynotes, Prezi, Mac/Windows operating system

Fluent in: Mandarin Chinese and English

EXPERIENCE:

Confidential, Ithaca, NY

Technical Consultant

Responsibilities:

• Designed a seasonal ticket holders’ attendance boost strategy based on historical data
• Constructed predicative models by analyzing 465,000 seasonal ticket holder data that include the holder’s attendance statistics, registration information and the games attended
• Developed the Binomial Logistic Regression Model that predicts the attendance for a given seasonal ticket holder and decide if a person needs to be incentivized
• Developed a Time Series Model that forecasts the general attendance for next 10 games; suggested that Spurs gives holders a promotion after the 4th game; suggestions accepted by client completely
• Led three Stats graduates; mainly contributed on model selection and creation of R and SAS code
• Presented finding and recommendations to client, Director of Graduate Studies, and faculties of MPS Programs

Confidential, Ithaca, NY

Responsibilities:

• Designed a transaction fraud prevention strategy based on historical data
• Developed a logistic regression model by analyzing 437,000 credit card transactions data that include detailed information for each transaction, the spending profiles of customers, and an indicator of fraud
• Methodologies Involved: Logistic Regression, K - Fold Cross-Validation, Detecting the threshold for approval of the transaction
• In charge of communication with advisors, variable & model selection and creation of R and SAS code

Confidential, Berkeley, CA

Responsibilities:

• Analyzed fMRI dataset with 11K variables and 1500 observations; constructed a model that reduced computation time by 95% from 600 min to 30 min with sacrificing little accuracy
• Tested over 10 statistical techniques, such as Lasso, Ridge, Boosted Decision Trees, K-nearest-neighbors, Generalized Linear Model, Linear/quadratic discriminant analysis
• Teamed with two Math undergraduates; mainly contributed on model selection and creation of R code

Confidential, Berkeley, CA

Responsibilities:

• Predicted the strength of solar flares
• Prediction had been proven correct
• Methodologies Involved: AR, MA, ARMA, ARIMA, GLM, Spectral Analysis, PACF, etc.