The project provides a complete end-to-end workflow for building a binary classifier in Python to recognize the risk of housing loan default. It includes methods like automated feature engineering for connecting relational databases, comparison of different classifiers on imbalanced data, and hyperparameter tuning using Bayesian optimization.

Stochastic AUC Maximization with Deep Neural Networks

Evaluating machine learning methods for detecting sleep arousal, bachelor thesis by Jacob Stachowicz and Anton Ivarsson (2019)

Face recognition, tackling three different "old-school" Computer Vision techniques - as part of the Biometrics System Concepts course @ KU Leuven

Predicting Supreme Court Decisions

Improving credit risk model using Machine learning techniques. We use a host of ml models and neural network to solve the issue.

Solve complex real-life problems with the simplicity of Keras

This problem is a typical Classification Machine Learning task. Building various classifiers by using the following Machine Learning models: Logistic Regression (LR), Decision Tree (DT), Random Forest (RF), XGBoost (XGB), Light GBM and Support Vector Machines with RBF kernel.

A machine-learning project on detection of fraudulent credit card transactions.

Spam Classifier using NLP — A comparative machine learning project that classifies SMS messages as Spam or Ham using Bag of Words, TF-IDF, and Word2Vec. Evaluation is based on AUC-ROC scores. Includes trained models, code notebook, and visualizations.

An end-to-end machine learning pipeline for predicting customer churn using a Kaggle telecom dataset. This project includes data loading, cleaning, preprocessing, feature encoding, and model training with hyperparameter tuning using RandomizedSearchCV. Evaluation is performed using classification metrics and AUC score.

An Analysis and Machine Learning model to understand employee retention and predict churn as part of the Google Advanced Data Analytics Certificate Capstone Project.

Creating ROC graph for every target on data

Involves analyzing user data to identify patterns and build predictive models that can forecast whether users are likely to stop using the application.

Performed feature engineering, cross-validation (5 fold) on baseline and cost-sensitive (accounting for class imbalance) Decision trees and Logistic Regression models and compared performance. Used appropriate performance metrics i.e., AUC ROC, Average Precision and Balanced Accuracy. Outperformed baseline model.

Performance calculation tool for Hateful Memes Challenge

Modelo predictivo de aprendizaje supervisado para detectar clientes en riesgo de cancelar cuenta en Beta Bank.

Keras-based Deep Learning to cleanse patient data and predict their chances of survival

The displayed project constitutes a Placement_prediction_model. Which can be used to predict the Probability of getting hired for a job with an accuracy of 83% on testing data. This model comprises of an ensemble of decision tree classifiers.