People / Students / Class of 2024
Ye Joon Han graduated from University of California, Berkeley with a BA in Statistics and Data Science. Throughout his undergraduate journey, Ye Joon took a diverse range of data science courses, including machine learning and data analytics, data inference and decisions, data structures, advanced programming in R, econometrics, time series, business analytics, and so on. To apply his knowledge and gain practical experience in data science applications, Ye Joon actively engaged in several data science projects and research. Ye Joon participated in his first data science project led by The Economist Intelligence Unit (EIU). Working with team members and a data scientist at EIU, Ye Joon developed a Python-based web scraping platform that collected online retail prices from various countries to construct key performance indicators for measuring inflation and correlating it with the consumer price index. Not only did this project enhance his Python skills, but it also deepened his knowledge in SQL. Another enriching experience was his involvement in a data science research project led by Harvard Library. The primary objective of the project was to develop the user-friendly web application called Longhand, which is a VR and NLP technology-based word cloud generator software that converts high-frequency words from raw texts into 3D images in a virtual environment, using Python, Blender, spaCy, Streamlit, Sketchfab, and WebXR. As part of his responsibilities, Ye Joon tested the integration of topic modeling output variables to accurately place models in the 3D scene. This project opened his eyes to the vast potential of VR and NLP as powerful analytics tools in the data science domain. Furthermore, Ye Joon participated in a data science project with STAT News, an online news outlet by Boston Globe Media. He developed a Machine Learning and NLP-driven approach, employing neural networks and text similarity, to establish a comprehensive database of conflicts of interest (COI) disclosed in academic papers archived in PubMed. He also utilized Named Entity Recognition in spaCy, BERT, and Azure for accurately identifying company and author names to investigate the relationship with respect to COI. This experience sparked his fascination with NLP's capabilities and its potential for precise information extraction. Further, Ye Joon was part of a research project led by Professor Adam Badawi at Berkeley Law, which received the Ribbon of Excellence award for the Data Science Discovery program at University of California, Berkeley. Ye Joon and his team members fine-tuned the FinBERT model to differentiate factual statements from opinion statements in the EDGAR database of public filings maintained by the Securities and Exchange Commission. They then trained the model on approximately 650 labeled sentences from SEC filings, achieving an accuracy of 0.93 on the test set. Ye Joon was thrilled to see how machine learning and NLP could empower financial investors with better-informed decisions. Ye Joon is excited to start his academic journey at Northwestern University as a MLDS student. He is looking forward to expanding his data science knowledge and expertise through collaboration with MU cohort students. In particular, he is enthusiastic about learning and exploring more about the applications of machine learning and artificial intelligence in various industry sectors. He believes that the MLDS program will facilitate his development as a competent data scientist.
