imbalanced-data-handling

Apply machine learning techniques to predict visa application approval outcomes based on applicant demographics, education, job experience, and employment-related attributes. The objective is to support more consistent, data-driven certification decisions by identifying key approval drivers and high-risk cases early in the process.

Showcasing impactful machine learning projects tackling real-world issues with Decision Trees, SVM, Time Series, Random Forest, Logistic Regression, and Gradient Boosting. Demonstrates advanced preprocessing, feature engineering, and optimization techniques.

Data-driven targeting of customers predicted to be at risk of canceling their credit card services

This project builds and compares ML models for fraud detection on imbalanced data using resampling, cost-sensitive learning, and ensembles. It includes a CLI for training and evaluation.

Credit Card Fraud Detection using Exploratory Data Analysis and Machine Learning to identify fraudulent transactions efficiently.

Deep learning project comparing CNN and MobileNetV2 for image classification on a small, imbalanced dataset. Includes preprocessing, augmentation, training, evaluation, and performance analysis using Python, TensorFlow, Keras, and Scikit-learn.

Deep learning | Built a transformer-based email classifier using BERT to detect spam with 93% accuracy, including preprocessing, class balancing, embeddings, model training, and evaluation.

Random Forest Classifier for Customer Churn Prediction

This project aims to identify fraudulent transactions using a real-world dataset containing credit card transaction data. Fraud detection is a highly imbalanced classification problem where fraudulent transactions are significantly rare compared to genuine ones.

ML project detecting fraudulent credit card transactions using Logistic Regression, Random Forest, and XGBoost. Handles imbalanced data with SMOTE and comprehensive evaluation metrics.

AI-powered system to detect fraudulent transactions in e-commerce using machine learning. Includes data preprocessing, feature engineering, and classification models like Random Forest and XGBoost. Achieved high accuracy with interpretable results for real-time detection.

Skin cancer classification using the HAM10000 dataset. Focused on handling highly imbalanced data with preprocessing, augmentation, transfer learning, and class-balanced evaluation metrics

Predicting customer booking completion for British Airways using Random Forest & XGBoost. Built as part of the Forage Data Science Virtual Job Simulation. Covers EDA, feature engineering, class imbalance handling, threshold tuning, and ROC-AUC model comparison.

"ML pipeline for customer churn prediction and risk segmentation using Python"