MLJAR AutoML
mljar.com
- Free plan
- Windows
- macOS
- Linux
- Web
- Self-hosted
What MLJAR AutoML is
MLJAR AutoML automates supervised machine learning for tabular data, covering binary and multiclass classification and regression. Users can run it in MLJAR Studio notebooks or use the open-source Python package in scripts and notebooks. Its four modes—Explain, Perform, Compete, and Optuna—provide configurable choices for algorithms, validation, metrics, time budget, and explanations. Depending on the mode and settings, workflows can preprocess data, select and tune models, engineer and select features, and create ensembles. Experiment outputs can include model comparisons, validation results, training times, explanations, and saved artifacts, with summaries in Markdown and JSON. Fairness-aware training evaluates models against a chosen metric and threshold and can try sample reweighting, but does not guarantee a fair model. The Python package integrates with Mercury for apps that make single or batch CSV predictions. Those apps can be previewed locally, published through MLJAR Platform, or self-hosted with Docker. The package is open source under the MIT license and costs nothing.
Who it suits
MLJAR AutoML suits people building supervised models from tabular data who want configurable workflows in notebooks or Python scripts. Its Mercury integration may also suit users who need apps for individual or batch CSV predictions.
Good to know, and what to watch
- Supports binary, multiclass, and regression tasks.
- Offers four configurable training modes.
- Open-source Python package is MIT licensed.
- Experiment summaries can be produced in Markdown and JSON.
- Mercury apps can be self-hosted with Docker.
- Designed for supervised tabular tasks.
- Fairness-aware training does not guarantee a fair model.
- Reports and explanations depend on task and configuration.
- No forecasting workflows.
Plans
Every plan and priceFree
mljar-supervised Python package
Pro
Business
Open source
Getting started
- Visit the MLJAR AutoML website.
- Choose MLJAR Studio notebooks or the open-source Python package.
- Run the package in scripts or notebooks and configure a training mode and settings.
- Use Mercury integration to create a single-prediction or batch CSV app.
- Preview the app locally, publish through MLJAR Platform, or self-host with Docker.
The verdict
MLJAR AutoML offers configurable automation for supervised tabular modeling, with experiment summaries and a path to prediction apps through Mercury. Its scope is limited to supervised tabular tasks, and its available explanations vary with settings.
Questions people ask
Is MLJAR AutoML free?
The mljar-supervised Python package is free and is open source under the MIT license.
What kinds of machine-learning tasks does it support?
It automates supervised tabular tasks including binary classification, multiclass classification, and regression. Forecasting workflows and time-series modeling are not supported.
What training modes are available?
The modes are Explain, Perform, Compete, and Optuna. Users can configure algorithms, validation, metrics, time budgets, and explanations.
Can it create prediction apps?
Yes. The Python package integrates with Mercury for single predictions and batch CSV predictions. Apps can be previewed locally, published through MLJAR Platform, or self-hosted with Docker.
Does fairness-aware training ensure fair results?
No. It evaluates models against a chosen metric and threshold and can try sample reweighting, but does not guarantee a fair model.
What platforms are listed?
Linux, macOS, self-hosted, web, and Windows are listed.
What the maker publishes
- Automated workflow
- AutoML can preprocess data, select and tune models, engineer and select features, and create ensembles when enabled by the selected mode and configuration. source
- Company
- MLJAR identifies Aleksandra Płońska as Founder and Executive Director and Piotr Płoński as Founder and CTO. source
- Deployment
- Generated Mercury apps can be previewed locally, published through MLJAR Platform, or self-hosted with Docker. source
- Fairness
- Fairness-aware training evaluates models against a chosen metric and threshold and can try mitigation through sample reweighting, but does not guarantee a fair model. source
- Headquarters
- Łapy, Poland source
- Integrations
- The Python package integrates with Mercury to generate prediction apps that support single predictions and batch CSV predictions. source
- Models and reports
- Experiments include model comparisons, validation results, training times, explanations, and reusable saved artifacts, with structured summaries available in Markdown and JSON. source
- Notable limits
- MLJAR AutoML is designed for supervised tabular tasks, and the available reports and explanations depend on the task, mode, algorithms, and explanation settings. source
- Purpose
- MLJAR AutoML automates supervised machine learning for tabular data, including binary classification, multiclass classification, and regression. source
- Security and privacy
- The privacy policy says MLJAR uses SSL/TLS encrypted connections, access controls, and authentication mechanisms, and advises users not to submit sensitive or regulated data to AI features unless authorized. source
- Support
- MLJAR’s About page invites users to contact the team with questions and says they are happy to help. source
- Training modes
- The four modes are Explain, Perform, Compete, and Optuna, with defaults that users can configure for algorithms, validation, metrics, time budget, and explanations. source
- Ways to use
- Users can run AutoML through MLJAR Studio notebooks or use the open-source Python package in scripts and notebooks. source
Checked against the maker's pages on 9 Oct 2026.