igel is a delightful open-source machine learning tool that lets you train, test and use models without writing code, by describing what you want in a simple configuration file or CLI command. Created by nidhaloff and released under the MIT license, it has gathered about 3,140 stars on GitHub and received updates into late 2025. It wraps scikit-learn’s algorithms behind a friendly interface aimed at people who have data and questions, but not necessarily a data-science background.
With igel you point the tool at a CSV dataset, declare the target column and the task type, and it handles preprocessing, model fitting and evaluation for you. It supports classification, regression and clustering across scikit-learn’s algorithm zoo, from random forests and gradient boosting to support vector machines and neural networks. Trained models can be saved and later used for predictions on new data through the same no-code workflow, and everything is configured through readable YAML or JSON files that are easy to version and share.
Typical igel workflows start from a CSV file: you write a short YAML configuration naming the target column, the model type and any preprocessing choices, then run one command to train and another to predict on new data. Behind the scenes igel handles train/test splitting, encoding and scaling, and reports standard evaluation metrics for the chosen task. This makes it handy for quick baselines before committing to a full modelling effort, for teaching machine-learning concepts without boilerplate, and for non-programmers who need a repeatable way to fit classical models to tabular data.
igel is free and open source under the MIT license. There are no paid tiers or hosted plans; it runs entirely on your own machine.
Analysts, students and developers who want to experiment with classical machine learning on tabular data quickly, and teachers looking for a gentle on-ramp to ML concepts.