A collection of applied data science projects focused on predictive modeling, segmentation, forecasting, and machine learning analysis to support data-driven decision-making.

 

Project Categories

Each project demonstrates applied machine learning and data science techniques across structured and unstructured data problems.

I. Predictive Modeling & Segmentation

Projects focused on classification, regression, clustering, and customer/business analytics.

Credit Risk Prediction Model

  • Built classification models using Logistic Regression, Random Forest, and XGBoost to predict loan default risk
  • Engineered financial risk features and applied SHAP for model interpretation
  • Evaluated model performance using ROC-AUC, precision, recall, and F1-score View on GitHub

Supply Chain Risk Prediction

  • Built classification models using Logistic Regression and XGBoost to predict supplier delay risk
  • Engineered supplier performance features to improve predictive performance

  • Benchmarked model performance to support supplier risk assessment and inventory planning

    View on GitHub

Customer Value & Lifecycle Modeling

  • Applied K-Means clustering and PCA for customer segmentation
  • Built an XGBoost regression model to estimate Customer Lifetime Value (CLV)

  • Designed an A/B testing simulation framework to evaluate customer retention strategies

    View on GitHub

House Price Prediction

  • Built regression models using linear and tree-based algorithms to predict housing prices
  • Performed feature engineering using property, geographic, and temporal variables

  • Compared model performance using cross-validation and regression metrics

    View on GitHub

 

II. Time Series & Predictive Analytics

Projects focused on forecasting, time-series analysis, and scenario-based decision support.

Financial Forecasting

  • Developed time-series forecasting models using Prophet, ARIMA, and regression techniques to forecast financial market trends
  • Engineered temporal features including lag variables, rolling statistics, and volatility indicators

  • Performed scenario analysis to evaluate potential future market outcomes

    View on GitHub

Healthcare Workforce

  • Built regression, Random Forest, and ARIMA models to forecast patient demand and estimate staffing requirements
  • Applied scenario analysis and Monte Carlo simulation to evaluate staffing needs under varying demand conditions

  • Designed a decision-support framework for workforce planning and resource allocation

    View on GitHub

Toronto Infection Outbreak Forecasting

  • Developed Prophet time-series forecasting models to analyze infection outbreak trends across healthcare facilities
  • Built interactive Power BI dashboards to visualize outbreak patterns and support public health decision-making

  • Applied scenario analysis to support healthcare resource planning under projected outbreak scenarios

    View on GitHub

 

III. Natural Language Processing & Generative AI

Projects involving transformer models, natural language processing (NLP), and generative AI applications.

Memory Support Chatbot (GPT-2)

  • Developed a GPT-2–based chatbot to provide context-aware support for pregnancy-related memory concerns
  • Applied NLP preprocessing and language model fine-tuning to improve response quality and relevance

  • Built interactive chatbot interfaces using Gradio and Streamlit for user interaction

    View on GitHub

Question Answering System (RoBERTa / SQuAD)

  • Developed a transformer-based question answering application using the RoBERTa SQuAD2 model
  • Built an interactive Gradio interface for real-time question answering

  • Implemented an inference workflow using Hugging Face Transformers to extract answers from user-provided context

    View on GitHub

Hugging Face Fine-Tuning QA System

  • Fine-tuned a Hugging Face transformer model on the SQuAD dataset for extractive question answering
  • Developed a Gradio application to interact with the fine-tuned model using custom context and questions

  • Demonstrated the end-to-end workflow from transformer fine-tuning to interactive inference using Gradio

    View on GitHub

 

IV. Data Analysis & Visualization

Food Preferences Streamlit App

  • Developed an interactive Streamlit application to explore food choice and nutrition survey data
  • Built interactive visualizations and filtering features to analyze demographic and behavioral patterns

  • Applied exploratory data analysis to communicate insights through an accessible, user-friendly interface

    View on GitHub

Obesity Data Analysis

  • Conducted exploratory data analysis on obesity, lifestyle, and demographic data to identify key behavioral patterns
  • Performed statistical analyses using ANOVA, Cramér’s V, and correlation analysis to evaluate relationships between health and lifestyle factors

  • Created accessible visualizations and communicated data-driven insights to support public health awareness and decision-making

    View on GitHub

✨ More projects available on my GitHub