EDUCATION

COLUMBIA UNIVERSITY, MAILMAN SCHOOL OF PUBLIC HEALTH

Expected May 2024

M.S. in Biostatistics, Data Science Track

Relevant Coursework: Data Science, Probability, Biostatistics

THE OHIO STATE UNIVERSITY

Aug 2018 - May 2022

B.S. in Biochemistry, Minors: Statistics, Pharmaceutical Science, Public Health, GPA: 3.8/4.0

Relevant Coursework: Statistical Modeling for Discovery, Bayesian Analysis and Statistical Decision Making

TECHNICAL SKILLS

Programming Languages: R, SAS, SQL, Python, Java, STATA, Scala

Tools: MATLAB, Tableau, PySpark, Sklearn, PyTorch, NumPy, Pandas, TensorFlow, MongoDB, IntelliJ IDEA, NLTK, Dataiku

WORK EXPERIENCES

NEW YORK PRESBYTERIAN HOSPITAL - COLUMBIA UNIVERSITY - Data Analyst

Oct 2022 – Present

  • Conduct Survey Data

ENCOMPASS STUDENT ORGANIZATION - OSU - Data Analyst Intern

Sep 2021 – May 2022

  • Built Tableau dashboard to visualize different projects’ progress (e.g.funds received), saving 5 hours per week of manual reporting work.
  • Aggregated unstructured data from 20+ sources and built Random Forest Model to predict drug misuse risk based on 10 characteristics(e.g. age, gender, race, living arrangement); targeted 10,000+ potentially at-risk individuals within two months.

THE OHIO STATE UNIVERSITY - Laboratory Assistant

Jun 2020 – Aug 2021

  • Performed P granules interaction network in Cytoscape from the PPI data to explore its properties and explain that P granule proteins formed a dense protein interaction network in JavaScript.
  • Visualized the relationship between 3’ shortening and 3’ addition of piRNA from experiment data with three Ph.D. Students and concluded that the activities of two regulators will suppress piRNA tailing in Python.

PROJECTS

HEART ATTACK ANALYSIS AND PREDICTION

Python Machine Learning Data Visualization EDA

  • Details：
  • Performed EDA to investigate the features’ relationship and extracted 10 high correlation variables (e.g. maximum heart rate, resting blood pressure, number of major vessels) to develop a good-fitting and parsimonious Heart Attack Prediction Model to filter 23% useless features.
  • Compared the accuracy scores and AUC from four algorithms to narrow down on better performing techniques, and selected Random Forest Classifier model with 93% accuracy scores and 90.3% AUC in prediction.
  • Improved prediction accuracy score by 2.45% through cleaning and transforming raw data to limit the effect of abnormal extreme values in the prediction model.

VACCINATION INVENTORY SYSTEM

Python NLP Numpy Pandas Seaborn Machine Learning Data Visualization EDA

  • Details：
  • Cleaned, visualized, and preprocessed the 2518 vaccination-related tweets dataset with NumPy, Pandas, and Seaborn to build an unbiased ML model and selected tweets text from 10+ variables as input, and verification status as output.
  • Performed NLP sentiment analysis on the text of the tweets by implying TensorFlow Keras with 0.9205 accurate scores in users’ vaccination status prediction for better preparation in further vaccination campaigns.

EARTHQUAKE PREDICTION

Python MongoDB PySpark Machine Learning Data Pipline

  • Details：
  • Extracted, transformed, and loaded raw data into MongoDB with PySpark to create decentralized storage systems that can analyze real-time data to meet the growing storage and computing needs of earthquake data.
  • Predicted earthquakes in 2017 based on the dataset from 1965 to 2016 through developing a random forest model pipeline with 89% accuracy scores and 0.402 RMSE.
  • Presented earthquake information by creating geo-map plot with a prediction via Bokeh and a frequency bar chart in a dashboard to easily view and manipulate predicted earthquake locations and magnitudes on the map.