data scientist

JINGFANG TANG

515-***-**** ********.***@*****.*** https://www.linkedin.com/in/jingfang-tang-05941750 SUMMARY

3+ years of experience with predictive model development, feature engineering, machine learning, and solid background in statistics.

Pro cient in Shiny, R and SAS, experienced in SQL, Tableau, Python, Matlab, and C.

Specialties: Deep learning, XGBoosting, logistic regression modeling, hypothesis testing, A/B testing, Bayesian analysis, generalized linear models, mixed models, time series analysis. EDUCATION

M.S. Statistics, Iowa State University September 2014-December 2016 M.S. Industrial Engineering, Stony Brook University August 2012-May 2014 RELEVANT EXPERIENCE

Master Creative Component Project

R jags Implementation of Bayesian Neural Network Fitting February 2016-November 2016 Major contributions: proposed and applied a Bayesian neural network approach for classi cation and regression using infertility and housing price data sets in R Shiny.

Built logistic regression models and used Bayesian approach for parameter tting.

Visualized model results and generated model output using R jags and Shiny.

Selected di erent statistical models for the modeling of the data sets and carried out parameter tuning to improve the predictive accuracy.

Developed an interactive web application using R and Shiny (https://github.com/gladystang). Kaggle Competitions

Santander Bank Customer Satisfaction Identi cation March 2016-May 2016 Won top 10% out of 5000+ teams. To predict the probability of unsatis ed customer for the data set which has 150,000+ cases, collaborated with team on problem solving, feature selection, and model tting.

Implemented data validity checking, data imputation and data cleaning. Created new variables and removing redundant and highly correlated variables for predictive models.

Utilized machine learning approaches (including PCA, LASSO, LDA, Best Subset and Ridge Regression) in feature engineering and applied advanced methods (including RandomForest, Ad- aBoosting, XGBoosting) for prediction.

Evaluated models by comparing performance of di erent predictive methods in terms of prediction accuracy (https://github.com/gladystang/Prediction-Santander-Bank-Customer-Satisfaction). Predicting Forest Cover Type in the Roosevelt National Forest March 2015-May 2015 Achieved top 2% in this contest. Major contribution to the team: performed data preparation, con- ducted exploratory data analysis, and utilized predictive methods (including RandomForest, ExtraTrees and Gradient Boosting) to classify the forest categories in R (https://github.com/gladystang/Forest- Cover-Type-Prediction).

SELECTED GRADUATE COURSES AND TEACHING COURSES

Selected Courses: Multivariate Statistical Learning, Data Mining, Machine Learning, Statistical Infer- ence, Statistical Methods (I,II,III), Time Series, Design of Experiments, Bayesian Statistic. Teaching Courses: Introduction to Business Statistics, Engineering Probability, Statistical Methods, Introduction to Statistics and Engineering Statistics.

Contact this candidate