Quantitative Research
Statistical Modeling · Machine Learning · NLP · Financial Analysis
Embedded Files
ECO225 Videoresearch_paper.pdfExplaining Police Killings: The Poverty and Crime Nexus
Explaining Police Killings: The Poverty and Crime Nexus
ECO225: Big-Data Tools for Economists (2025-2026 Winter)
ECO225: Big-Data Tools for Economists (2025-2026 Winter)
Constructed large-scale datasets with 22,000+ observations for exploratory and quantitative analysis, producing statistical and geographic visualizations
Sourced, merged, and cleaned public datasets using Python (Pandas, NumPy), with additional data collected through API extraction and web scraping (BS4)
Created new metrics and conducted data visualization, descriptive statistics, and OLS/multiple regression analysis
Applied Lasso, Ridge, Elastic Net, clustering, Random Forest, and XGBoost with hyperparameter tuning
Solo-authored the research paper in Overleaf with LaTeX and used GitHub for version control
Skills: Pandas, Numpy, BS4, OLS, Lasso, Ridge Regressions, Clustering, Random Forests, XGBoosting, API Data Extraction, Web Scraping, GitHub Version Control
Classification & Ensemble Methods for Loan Default Prediction
Classification & Ensemble Methods for Loan Default Prediction
RSM338: Applications of Machine Learning in Finance (2025-2026 Winter)
RSM338: Applications of Machine Learning in Finance (2025-2026 Winter)
Analyzed 12,700+ Lending Club loan records to identify key borrower and loan characteristics associated with default risk
Cleaned and transformed data by addressing missing values, encoding categorical variables, and standardizing features for machine learning models
Built and compared seven classification models, including Logistic Regression, Random Forest, AdaBoost, and XGBoost, using cross-validation and GridSearchCV for model optimization
Evaluated models using accuracy, precision, recall, confusion matrices, ROC/AUC, and training time; XGBoost achieved the strongest performance with 77.71% test accuracy
Analyzed feature importance and optimized classification thresholds based on the financial costs of false predictions, recommending a 0.3 threshold for XGBoost
Skills: Python, Pandas, NumPy, Scikit-learn, XGBoost, Cross-Validation, GridSearchCV, Data Visualization, ROC/AUC Analysis
rsm338 classification & ensemble methods.pdf
DuPont Analysis: Investigating Company Performances Across Industries
DuPont Analysis: Investigating Company Performances Across Industries
RSM326H1: Data Analytics with Financial Accounting Information (2025-2026 Fall)
RSM326H1: Data Analytics with Financial Accounting Information (2025-2026 Fall)
Analyzed financial statement data for 170+ public companies, evaluating profitability, efficiency, leverage, and return on equity across industries using the DuPont framework
Built interactive dashboards to compare industry and company performance, identifying trends in financial ratios and changes over time
Skills: Excel (Power BI, PivotTables, XLookup), Tableau, financial statement analysis, DuPont analysis
Portfolio Optimization Analysis
Portfolio Optimization Analysis
ECO204Y1: Microeconomic Theory and Applications (for Commerce) (2024-2025)
ECO204Y1: Microeconomic Theory and Applications (for Commerce) (2024-2025)
Analyzed monthly financial data (Sep, 1981 - Oct, 2024) for two equities, evaluating return, risk, and performance using descriptive statistics and risk-adjusted metrics
Built and compared four portfolio optimization models (Statistical, Shrinkage, Wiener, and Itô) to determine optimal asset allocations
Skills: Excel (XLOOKUP, VLOOKUP, Data Analysis ToolPak), portfolio optimization, financial analysis, descriptive statistics, risk analysis
Leung_HoiYingVenice_09926.xlsxPage updated
Google Sites
Report abuse