A selection of my work demonstrating expertise in key data science techniques:

Each project showcases how I apply these analytical techniques to solve real-world problems.

 

I. Predictive Modeling

These projects showcase predictive modeling using algorithms like XGBoost, Random Forest, and regression to predict outcomes, quantify risk, and guide data-driven decision-making.

Credit Risk Prediction Model

  • Problem: Predict credit risk to help financial institutions make informed lending decisions.
  • Approach: Applied XGBoost and Random Forest with feature engineering and model evaluation.
  • Technologies: Python, XGBoost, Random Forest, Azure, Power BI
  • Results: Achieved high accuracy in identifying high-risk customers, enabling better lending decisions.
    View on GitHub

Financial Customer Retention Analytics

  • Problem: Predict customer churn and estimate lifetime value (CLV) for financial services.
  • Approach: Gradient boosting models on transactional/demographic data integrated into dashboards.
  • Technologies: Python, Gradient Boosting, SQL, Power BI, Azure ML
  • Results: Provided actionable insights for improving customer retention and maximizing lifetime value.
    View on GitHub

Predictive HR / Workforce Analytics

  • Problem: Predict employee attrition to improve retention strategies.
  • Approach: Random Forest and XGBoost on HR datasets; interactive dashboards for retention insights.
  • Technologies: Python, XGBoost, Random Forest, SQL, Power BI, Plotly Dash
  • Results: Identified high-risk employees and guided HR teams to implement effective retention strategies.
    View on GitHub

 

II. Time Series Forecasting & Simulation

These projects demonstrate modeling of time-dependent data and using simulations to predict resource demand, trends, and scenario outcomes.

Healthcare Resource Forecasting

  • Problem: Forecast patient volume and resource demand for better hospital planning.
  • Approach: ARIMA + Monte Carlo simulations; automated retraining pipelines in Azure ML.
  • Technologies: Python, ARIMA, Monte Carlo, SQL, Plotly Dash, Azure ML
  • Results: Enabled hospital administrators to anticipate resource needs and allocate staff efficiently.
    View on GitHub

Healthcare Workforce & Scheduling Optimization

  • Problem: Predict staffing requirements and evaluate scheduling interventions.
  • Approach: Regression and Random Forest; statistical analysis using A/B testing and bootstrap.
  • Technologies: Python, Random Forest, Regression, SQL, Plotly Dash, Tableau, Azure ML
  • Results: Improved staffing efficiency and optimized schedules based on predictive insights.
    View on GitHub

Epidemiology: Toronto Infection Outbreaks (2016–2025)

  • Problem: Predict infection outbreaks and identify hotspots.
  • Approach: Prophet models for time-series forecasting; dashboards in Power BI.
  • Technologies: Python, Prophet, SQL, Power BI
  • Results: Provided actionable visualizations for monitoring infection trends and informing public health decisions.
    View on GitHub

Financial Forecasting & Scenario Analysis

  • Problem: Forecast financial KPIs and run “what-if” investment scenarios.
  • Approach: Hybrid Prophet + ARIMA; Monte Carlo simulations; interactive Plotly Dash dashboards.
  • Technologies: Python, Prophet, ARIMA, Monte Carlo, Plotly Dash
  • Results: Allowed dynamic scenario analysis for strategic financial planning and investment evaluation.
    View on GitHub

 

III. Data Visualization, Business Intelligence & App Deployment

These projects turn complex data into actionable insights via dashboards, apps, and automated workflows.

Operations Efficiency Dashboard

  • Problem: Monitor and improve operational KPIs.
  • Approach: Data cleaning and automated dashboards using Python, SQL, Power BI, and Azure Pipelines.
  • Results: Streamlined reporting, enabling faster decision-making and bottleneck identification.
    View on GitHub

Marketing A/B Testing Simulator

  • Problem: Simulate marketing campaigns and analyze impact.
  • Approach: Statistical tests (t-tests, ANOVA, bootstrap); automated reporting from Jupyter to HTML/PDF dashboards.
  • Results: Provided actionable insights for marketing strategy optimization and campaign evaluation.
    View on GitHub

Food Choices Streamlit App

  • Problem: Explore college students’ food preferences with interactive analysis.
  • Approach: Built a Streamlit app to collect, analyze, and visualize data.
  • Results: Enabled users to interactively explore nutritional trends and preferences.
    View on GitHub

 

✨ For more projects, visit my GitHub repositories