mljar-supervised

Why we included this project

mljar-supervised handles the whole model-building workflow for tabular classification and regression: preprocessing, feature engineering, trying algorithms like LightGBM, XGBoost, CatBoost, and neural nets, then tuning hyperparameters and combining results with ensembling and stacking. A single fit call can take you from raw CSV to a tuned model, which is useful for teams that would rather spend hours on the business problem than on pipeline plumbing. What sets it apart is transparency: each trained model ships with a detailed Markdown report covering feature importance and SHAP explanations, so results can be audited and explained to stakeholders. It also generates a deployable web app around a trained model, which helps people who do not want to write Python. That combination of automation and explainability makes it a practical choice when you need defensible models, not just predictions.