Yingzi Xu bio photo

Yingzi Xu

Ph.D. Student in Statistics @ NC State University

Email LinkedIn Github

Projects

Overview

Spatial Study of Gasoline Price Trend in United States

  • Visualized price trend and pattern difference of seven regions in U.S.

  • Developed a predictive model for future gas price in different regions of United State via time series analysis and spatial panel anlysis

  • Poster available here

Recommender System for Yelp Business Rating Prediction

  • Implemented different recommendation system models such as collaborative filtering model to predict the future ratings of business given by specific users given a detailed snapshot of Yelp data

Click-through Rate Prediction of Online Advertisements

  • Predicted the click-through rate of ads via multivariate linear regression models, logistic regression models and Support Vector Machine based on 16 million records of user information

Top 10 Articles in Wikipedia via Hadoop MapReduce

  • Explored a 41GB of XML text file in Wiki markup language to get the total word count and top 10 popular articles through Pig and MapReduce scripts in Python

A Bayesian Approach to Compare Survival Rate of Bone Marrow Transplant Methods

  • Developed both parametric and semi-parametric Bayesian Cox Proportional Hazards models to compare the effectiveness of two bone marrow transplant methods

  • Poster available here

Longitudinal Study of Different Tillage Treatments Effect on Soil Moisture

  • Consulting project for client Dr. Jeffrey G. White in Department of Soil Sciense in NCSU

  • Determined tillage effects on soil moisture while capturing trends and patterns by missing data handling, time series analysis, longitudinal analysis

  • Report available here
  • Slides available here

Longitudinal Study of the African American English Development

  • Developed longitudinal models to examine the evolution of the African American English vernacular, while accounting for time dependent / independent covariates

  • Investigated variance structure of repeated measurements, completed population level model, random coeffient model, linear mixed model and generalized linear mixed model

  • Report available here

Influence of Incentive Plans to Credit Card Consumption

  • Conducted questionnaire design, investigations and collected data

  • Discovered key factors that affect the credit card consumption by linear mixed mdel via SAS

  • Prepared professional research report to banks