My actual name is Hongyang, but you can also call me Alex. I am a data science fresh-graduate looking to work in the gaming industry. My main interest is to find ways to integrate machine learning technologies into games on a development level. I have extensive education background in mathematics, statistics, and data science.

Bachelor of Arts in Mathematics
University of Washington
2016 - 2020

• Relevant coursework: Function Analysis, Topology, Modern Algebra, Probability

Master of Science in Statistics
University of Haifa
2021 - 2022

• Relevant coursework: Statistical Learning, Mathematical Statistics, Markov Chain, Regression

Master of Data Science
University of British Columbia
2023 - 2024

• Capstone project: Sentiment Analysis Model Development
• Relevant coursework: Supervised and Unsupervised Machine Learning, Regression, Inference, Causal Analysis, Development Workflow, Visualizations

This repository is for a data analysis project on red wine quality prediction using different machine learning model. The project itself is nothing sophisticated as the primary focus was on creating a reproducible analysis report.

My specific contributions were:

• Fine-tuned parameters of different machine learning models for classification (Logistic Regression, SVM RBF, kNN, Decision Tree)
• Wrote multiple test functions for different functionalities
• Helped make the report reproducible using jupyter book
• Set up proper instructions for reproducing the report under virtual environment

This repository is for a visualization dashboard of a world happiness level data. The dashboard is interactive and allows the user to see trends and rankings based on user selections.

My specific contributions were:

• Developed dash module for a interactive pandas table in the dashboard
• Helped track different changes implemented as reflection

This repository is for our capstone project for the MDS program at UBC. We wanted to pave way for an even bigger model that can help determine a potential customer interested in buying a real estate listing as well as recommend different estates based on customer demands. Due to time constraint, we could not do everything. We focused on using fine-tuned and zero-shot models to predict customer interest level (labelled as three levels: cold, warm, and hot) with agent-customer text history as our data set.

My specific contributions were:

• Cleaned up and aggregated conversation log data set for compatibility with models
• Engineered features indicating agent response time and frequency for modelling
• Restructured the repository and modularized python codes from notebooks into scripts for reproducibility