Research

Research

Air Pollution Prediction from Acoustic Signals

Question

Can acoustic signals predict air-pollution levels?

My role

Integrated time-series datasets, engineered predictive features, and evaluated XGBoost models.

Methods

• time-series
• xgboost
• feature-engineering
• python
• r

Outputs

• Integrated 10GB+ of data from 30+ structured and semi-structured files in R.
• Explored acoustic–pollution relationships using nine correlation heatmaps and engineered 27 features.
• Achieved R² = 0.87 with XGBoost and k-fold cross-validation, outperforming Random Forest.

Paid · Research assistant

Satellite Portfolio Strategies

Question

Which satellite portfolio strategies are high-value?

Department

Department of Economics, University of Toronto

My role

Data collection using scraping and APIs in R, schema design, and analysis.

Methods

• regression
• difference-in-differences
• r
• sql
• web-scraping

Outputs

• Consolidated a 7-table SQL schema into 5 tables.
• Collected a dataset of 3,000+ rows.
• Difference-in-differences analysis and visualizations of satellite portfolio strategies.

↳ Also in Work