Machine Learning Projects

End-to-end machine learning systems built for prediction, automation, and real-world impact across environmental and healthcare domains.

CO₂ Emissions Prediction Model

Supervised regression models predicting vehicle CO₂ emissions using engineered features, exploratory analysis, and ensemble learning.

Python · Regression · Gradient Boosting · Data Science

Overview

This project builds a full machine learning pipeline to predict vehicle CO₂ emissions using Canadian government data. It spans data cleaning, feature engineering, modeling, evaluation, and real-world interpretation.

Process Highlights

Exploratory Data Analysis with correlation analysis
One-hot encoding for categorical features
Model comparison across Linear, Ridge, Lasso, ElasticNet
Gradient Boosting as top performer

Performance

R²: 0.995
MSE: 17.36

Business & Policy Relevance

Enables emission forecasting for new vehicle designs, supports environmental policy modeling, and informs consumer decisions.

⭐ View on GitHub

Python (pandas, numpy)
scikit-learn
matplotlib, seaborn, missingno
Regression & ensemble learning

Prediction of Cancer Types Using Gene Expression

Machine learning classifiers trained on RNA-seq gene expression data to identify cancer types with near-perfect accuracy.

Python · Bioinformatics · Classification · XGBoost

Problem Statement

Early cancer detection is critical. This project uses gene expression profiles to classify cancer types using supervised learning.

Dataset

RNA-seq expression data sourced from the UCI Machine Learning Repository, covering five cancer types including BRCA, KIRC, and LUAD.

Models Used

Logistic Regression
Support Vector Classifier
XGBoost

Results

Logistic Regression: F1 = 1.00
XGBoost: F1 = 0.9938

Impact

Demonstrates how machine learning can support early diagnosis, personalized treatment planning, and medical research.

⭐ View on GitHub

Python, scikit-learn
StandardScaler & Pipelines
RandomizedSearchCV
XGBoost