Hi, I'm Izan
I'm interested in the foundations of Data Analytics & Science
IK

About

Fourth-year Mathematics & Statistics student at McMaster University with a focus on data analytics and data science. Currently interning as an Analytics Intern at Tolmar Inc., a pharmaceutical company, where I apply analytical skills in a real-world industry setting. I enjoy turning raw data into clear, actionable insights through Python, SQL, and Power BI.

Work Experience

Volunteer Experience

Skills

Data Analytics
Tableau
SQL
Python
R
Data Visualization
Power BI
Machine Learning
Git
Microsoft Excel
Google Cloud Platform
ArcGIS Pro
ArcGIS Server
Microsoft Power Query
Microsoft Azure
Azure Synapse Analytics
My Projects

Check out my latest work

I've worked on a variety of projects. Here are a few of my favourites.

Bank of Canada Credit Card Arrears Time-Series Analysis

Tested whether changes in the 5-year mortgage rate predict changes in credit card arrears three quarters later. Pulled Bank of Canada data through the Valet API, built a 45-quarter dataset (2015–2026), tested the series for stationarity, and fit an OLS regression on differenced data. Found no significant link (n = 41, p = 0.36, R² = 0.02), even though the two series correlate at about 0.65 in levels, a reminder that shared trends can look like a relationship.

Python
pandas
statsmodels
Valet API

Multiple Binary Logistic Regression: UC Berkeley Admissions

Modeled admission odds by gender and department on the UC Berkeley admissions dataset using a frequency-weighted binomial GLM in R, reporting odds ratios with 95% confidence intervals. Found gender was not a significant predictor once department was controlled for, while Departments C to F had significantly lower odds of admission than Department A (p < 0.001). Visualized predicted probabilities by department and gender with ggplot2.

R
ggplot2

Credit Card Fraud Detection

Built a Random Forest classifier to detect fraudulent transactions across 284,807 records, achieving an AUC-ROC of 0.9731. Addressed severe class imbalance (0.17% fraud) using SMOTE, improving detection of the minority class. Evaluated performance using precision (0.85), recall (0.84), and AUC-ROC, correctly catching 82 of 98 fraud cases in the test set.

Python
scikit-learn
imbalanced-learn
Seaborn

Image Classification with Pre-Trained ResNet50

Built an image classifier using transfer learning with a pre-trained ResNet50 model, capable of classifying images into 1,000 object categories with top-5 confidence scores. Implemented preprocessing, evaluation, and deployment-ready inference logic with a simple web interface for demo use.

Python
PyTorch
torchvision

Ontario Hotel Performance EDA (2020–2023)

Analyzed occupancy, ADR, and RevPAR trends across 20+ Ontario regions using Ontario open data, finding Downtown Toronto had the largest COVID occupancy recovery (48.4 pp) from 2020–2023. Found a weak positive correlation (0.36) between occupancy rate and ADR, suggesting pricing and demand are largely independent across regional markets.

Python
pandas
Matplotlib
Seaborn

Multi-Department Revenue Analysis & Dashboard

Queried and cleaned 6 years of revenue data (2017–2022) from a public Kaggle department-store dataset using SQL, identifying a peak revenue year in 2018 followed by consistent year-over-year decline. Built an interactive Power BI dashboard visualizing department-level revenue trends for cross-unit performance comparison.

SQL
Power BI
GitHub
LinkedIn