Welcome!

Nutrient Composition of Common Foods in Canada: Analyzing the Canadian Nutrient File

Tech Stack:

Interactive dietary tools leveraging the Canadian Nutrient File — a Tableau dashboard for in-depth nutrient comparisons and a Streamlit app featuring dynamic visualizations, clustering analytics, and an AI chatbot for personalized nutrition guidance. Presented at YYC DataCon 2025.

AF Risk Prediction Using ECG & EHR Data

Tech Stack:

An end-to-end pipeline and interactive app for forecasting new-onset atrial fibrillation using routinely collected 12-lead ECG signals and EHR data. The XGBoost model identifies high-risk patients, while the Streamlit interface and DeepSeek chatbot provide intuitive exploration and insights. Presented at Statistical Society of Canada Annual Meeting 2025.

Credit Card Default

Tech Stack:

A predictive modelling project to forecast credit card defaults using demographic and financial data from 30,000 account holders — combining exploratory data analysis, visualization, and machine learning (regularized classifiers and XGBoost) for robust risk prediction.

Enhancing Bank's Personal Loan Approaches

Tech Stack:

A sequential neural network model predicting customer likelihood to accept personal loan offers — achieving 98% accuracy, 96% precision, and 87% recall — to drive targeted marketing strategies.

Sleep-Apnea Severity Estimation from EEG Spectrograms via ResNet-18

Tech Stack:

An end-to-end deep learning pipeline that regresses continuous AHI from EEG spectrograms using a ResNet-18 backbone. With subject-stratified validation, it achieves an RMSE of 6.8 events/hour and a Pearson correlation coefficient of 0.76, delivering accurate severity rankings across the entire AHI range. The workflow combines robust preprocessing, transfer learning optimization, and scalable inference for real-time sleep apnea screening.

GDP & Population: 2024 Olympic Medals

Tech Stack:

An analysis of how GDP, population, and socio-economic indicators relate to medal performance at the 2024 Olympics — using correlation matrices, regression models, and clustering techniques to uncover key drivers of success.

BeWell360: Your Daily Holistic Wellness Log

In Progress

BeWell360 is a holistic wellness tracker that uses data science and analytics to transform daily logs — nutrition, fitness, sleep, body composition, professional development, and personal growth — into actionable insights for balanced living.

Database Chatbots

Tech Stack:

AI-powered agents for conversational interaction with relational data — transforming natural language prompts into SQL queries and insights for data-driven decision-making.

Designing an Intelligent Agent for the Wumpus World

Tech Stack:

An intelligent Wumpus World agent that navigates a grid of hazards and searches for gold under uncertainty. It perceives environmental cues — breezes near pits and stenches near the Wumpus — applies logical inference to deduce safe paths, and balances exploration with caution to avoid fatal encounters, demonstrating key AI concepts such as decision-making under uncertainty, logical reasoning, and informed search strategies.

YT Channel Analytics

Tech Stack:

A Streamlit app for analyzing YouTube channel performance — tracking subscriber growth, views, watch hours, likes, comments, and shares to uncover engagement insights.

Wine Dataset Analysis

Tech Stack:

An exploratory analysis of red and white wine quality — examining the proportion of high-quality wines, identifying key differentiators (e.g., alcohol content, sulphur dioxide), and evaluating their impact on ratings.

Unlocking Superconductor Potential: Predicting Critical Temperatures with Multiple Regression

Tech Stack:

A multiple regression analysis to predict superconductors' critical temperature (Tc), identify the most impactful material features, and evaluate model performance.

Life Expectancy EDA: Key Influencing Factors

Tech Stack:

A focused exploratory analysis of life expectancy drivers — examining healthcare investment, vaccination coverage, and socio-economic indicators — and contrasting trends across developed and developing countries.

LinkedIn and Mental Health: UCalgary MDSA Students

Tech Stack:

An interactive dashboard examining how LinkedIn engagement influences the mental health of UCalgary MDSA students — developed from a carefully designed survey to test hypotheses such as whether exposure to peers' success posts heightens anxiety or whether supportive comments enhance job-search confidence. The dashboard presents trends in confidence, anxiety, and comparison behaviours using simplified versions of the original survey questions, which were more comprehensive and formally structured — for example, "How motivated or anxious do you feel after browsing LinkedIn?" and "How often do you compare your progress to others online?" Users can explore correlations, observe sentiment shifts, and filter results by gender, age range, student status, and program specialization to reveal demographic patterns and identify areas where targeted support may be most beneficial.

LEGO Set Analysis: Patterns in Build Complexity, Price Points, and Themes

Tech Stack:

An interactive visualization of the LEGO universe, analyzing 4,385 sets to reveal trends in average pieces, prices, and themes across age ranges. Users can filter by theme group, specific theme, or target age range to examine how complexity and pricing vary across categories. Key metrics — including total sets, average piece count, and average price — enable comparisons between themes, identification of high-value categories, and detection of design and pricing trends over time. Designed for collectors, enthusiasts, and data analysts, it offers a clear and engaging way to explore LEGO's evolution through data.

Analyzing 2024 Olympic Medals: The Role of GDP and Population

This article explores the relationship between a country's population size and GDP per capita and its success in the 2024 Olympic Games. By analyzing these variables, we aim to understand how they influence the total number of medals won by different countries. The article also involves clustering countries based on their Olympic performance and socioeconomic context, providing insights into patterns and trends affecting their athletic achievements. Through this analysis, we seek to uncover the extent to which economic and demographic factors contribute to Olympic success.

Understanding the P-Value in Hypothesis Testing

In hypothesis testing, we start with the null hypothesis (H₀), assuming no effect or difference. We then collect our observed results — the actual data from the study — and calculate the p-value, which tells us how surprising these results are if H₀ is true. A small p-value (e.g., < 0.05) leads us to reject H₀, considering the alternative hypothesis (H₁) as a possible explanation. A large p-value suggests our results align with H₀, so we fail to reject it. Importantly, the p-value is not the probability that H₀ is true; instead, it's the probability of observing our results, or more extreme ones, under the assumption that H₀ is correct.

Essential DAX Functions for Power BI: The Story Behind the Code

This article is a clear, practical guide to DAX in Power BI, focusing on understanding what you want to calculate rather than memorizing formulas. It explains how to choose between measures and calculated columns, how row and filter context work, and how to use key functions for totals, logic, and time-based calculations. With simple examples of CALCULATE, SUMX, IF, SWITCH, and more, it also covers text formatting, selection tools, and ways to filter or rank data. Tips on using VAR/RETURN help make formulas easier to read and faster to run, encouraging you to see DAX as both a language and a tool for building powerful reports.