This python code is an individual work and also my thesis work which contains a loss function with penalty to solve the problem of imbalanced classes and also the simple ANN to detect if a transaction is fraudulent or not in the case of credit card fraud. So with this algorithm, we can easily detect fraudulent transaction with a good precision .

2017 repository

Predicting vessel transshipment (discharge) amounts using XGBoost regression trees on a very small dataset.

Basic POC to showcase how to sample data for an imbalanced ad-clicks predictive model.

Classification on an imbalanced dataset, evaluating several model-resampling method combinations with hyperparameter tuning.

Implementation of semi and self supervised learning on Imbalanced Dataset

Implementation of C4.5 + Binarization (OVO / OVA) with/without SMOTE preprocessing. This way, multi-class imbalanced problems can be addressed

We use the widely available Kiva Dataset to predict if a Kiva Loan posting will get funded or not

Predicting credit card defaults (Classification)

LightGBM no-show predictor + AWS deployment (EC2 + Elastic Beanstalk)

Quick notebook to refer to different ways to handle imbalanced datasets.