The platform helps you focus on solving problems with data science

Experiment with AI management platform


Building a model is an iterative process that can take weeks, months, or even years, and reproducing model results, maintaining version control, and auditing past work is complex.


experiment tracking automl

Experiment tracking

Track and compare model performance across different technologies and approaches.

  • Why? Model building is an iterative process. Ideally you record not only each step, but also how you arrived there. A model shouldn’t be a file hidden away somewhere, but instead a tangible object that all parties can track and analyze consistently
  • How? allows you to record each experiment as you train it along with its characteristics, automated analyses, and versions as your project progresses, whether you created it using our AutoML or your own tools.


Automatically experiment with dozens of feature engineering strategies and algorithm types to build highly performant models.

  • Why? There are no silver bullets in Machine Learning, and you can’t be sure of the best solution until you’ve experimented with most of them.
  • How? In a single command, the engine automatically tries out different feature engineering strategies for every type of data (e.g. tabular, text, images) to maximize the information in your datasets.
AutoML capacities for experiment

Version Comparison

Compare various trained models to understand better the impact of train parameters on performance and decide which model is the most suitable for your use case.

  • Why? Because experiments can generate many models and iterations and the ability to track and compare the results is key to reliably choose the best model for deploying into production.
  • How? By comparing side by side training parameters and train results between two models. Train parameters and interactive performance charts are displayed in a single page.

Automated analyses & reports

Get deep insight in seconds about the behavior of your models.

  • Why? Models need to be analyzed in depth to be understood, and those analyses need to be accessible to all parties that have an interest in the model lifecycle.
  • How? automatically creates a full analysis dashboard for all models imported into the platform and provides convenient methods for further analysis using the SDK.
automated report for experiment
compatibility R Python and Node for experiment

Full compatibility between UI tools and Python, Node, and R SDKs

Click or code your way through projects, whatever is your preferred way to work.

  • Why? Data scientists come in many shapes, some from a more statistical background, other from a software engineering one.
  • How? All of Prevision’s features can be accessed through the User Interface, or via the Python and R SDKs. For users of additional languages, the platform also features a rich and documented API.