RapidMiner Automated Model Ops
Model deployment and management made easy – for any model and any user
Code free and process-free deployment
- Deploy models without the need for technical skills
- Manage multiple deployment locations and share deployed models with other users
- Drive collaboration on models, monitor for governance, drift and bias issues and even set up alerts and integrations
See how models work in the wild and adjust your strategy accordingly
- Deploy multiple models together where one model is active and others are ‘challengers’, where predictions and error rates are also stored for challengers
- Review performance of active and challenger models over time on the leaderboard
- Setup alerts to notify you when a challenger outperforms the active model
- Swap the active model with a single click so you’ve always got the best model deployed into your workflow
Performance & Business Impact
Go beyond ‘accuracy’ stats to understand and demonstrate financial impact of your models
- Show cumulative gains and business impact produced by your active model
- Analyze scoring times to pick models fast enough for your needs
- Compare distribution differences between predictions and actual values
Prevention of Model Drift & Bias
Instantly understand problematic trends and address them proactively
- Understand models and see how they behave with weights, model visualizations, interactive simulators and full prediction explanations
- Compare models’ latest performance with expected error rates to detect concept drift or shift
- Calculate input factor drift and compare drift with factor importance as early indicators for problematic concept shift or drift
- Observe suspicious changes between training and scoring data and detect bias in training data
RapidMiner Model Ops is part of a path to fully automated data science, from data exploration to modeling to production, when combined with Turbo Prep and Auto Model in RapidMiner Studio Enterprise.
See it in Action
Related Resources. Take a Look!
Read this 2018 Gartner Research Report detailing the challanges organizations face when deploying machine learning models into production.
In this webinar, Ingo dissects the issues that plague organizations striving to become ‘more AI-driven’ and prevent them from executing projects that have the potential to deliver incredible returns.