dsbox-ta2

DSBox is an AutoML system designed to automate the creation of comprehensive machine learning pipelines, encompassing data augmentation, cleaning, featurization, model selection, and hyperparameter optimization. It features a customizable pipeline template system that allows advanced users to tailor the pipeline search space while efficiently generating high-quality machine learning models with minimal intervention for non-expert users. The system is dockerized for easy deployment and includes support for various input/output formats consistent with D3M projects, producing JSON pipeline files for predictions.