Official Code for "Confidence Matters: Enhancing Medical Image Classification Through Uncertainty-Driven Contrastive Self-distillation" accepted at MICCAI2024

Credit card fraud detection system using logistic regression, EDA, class imbalance analysis, SMOTE, and a Gradio demo.

Customer Churn Prediction using Machine Learning (Imbalanced Classification) Customer Churn Prediction using Machine Learning (Imbalanced Classification)

Fundamentals of Machine Learning Assignment Repository

Predicting which people would be likely to convert from free users to premium subscribers in the next 6 month period, if they are targeted by our promotional campaign.

Predicting company bankruptcy using various machine learning models. The dataset is sourced from Kaggle: Company Bankruptcy Prediction.

(WIP): 'Aporia' in Greek means 'inconsistent'. A Python library that detects and fixes dataset issues using both rule-based methods and ML models. It evaluates dataset quality across multiple metrics, including missing values, duplicates, outliers, class imbalance, and label consistency. It also suggests fixes based on the metric scores.

Developing a machine learning model to predict customer churn as it is essential for proactively retaining valuable customers.

Comparing SMOTE, Class weight and Hybrid techniques for stroke prediction on an imbalanced medical dataset using Logistic Regression and Random Forest.

End-to-end credit risk scoring system using XGBoost, SHAP, and threshold tuning to predict loan defaults and automate lending decisions.

End-to-end machine learning workflow on the Combined Cycle Power Plant dataset: data cleaning, EDA, outlier removal, feature engineering, class balancing, and model evaluation for regression and classification. Includes code, visualizations and best practices in a single Jupyter notebook.

Supervised Learning project from TripleTen

Developed an ensemble ML classification model to predict U.S. visa case outcomes (Certified vs Denied) using applicant and employer attributes. Performed EDA, sampling, and model tuning (Random Forest, Gradient Boosting, XGBoost) to improve decision efficiency and identify key policy drivers like education, experience, and wage trends.

This project focuses on detecting fraudulent credit card transactions using Machine Learning and Data Analytics. It applies advanced techniques such as EDA (Exploratory Data Analysis), feature engineering, and imbalance handling (SMOTE, undersampling) to improve fraud detection accuracy.

This project builds an end-to-end machine learning pipeline to predict customer churn using the Telco dataset. It applies real-world data preprocessing, feature engineering, and multiple ML models with recall optimization for business impact. The final system is production-ready with model serialization using Joblib for deployment.

A binary classification task performed with machine learning in Python. The dataset's target distribution is heavily imbalanced. The model performance was evaluated with F1 scores.

This is a production-ready, end-to-end system developed to detect and classify racist tweets using advanced Natural Language Processing (NLP) techniques. Built on top of BERTweet (vinai/bertweet-base) and fine-tuned with a robust, k-fold cross-validation training pipeline, powered by streamlit UI!

Analysis of bank marketing campaigns using machine learning to predict term deposit subscriptions, optimizing campaign strategies through comparative evaluation of classification models.

Machine learning experiments analyzing crimes in Boston from real datasets and predict crime severity based on selected features.