Responsible AI In Practice
This blog series covers what happens when AI systems make a mistake, and why that matters.
Read the SeriesDataiku provides critical capabilities for explainable AI, including interactive reports for feature importance, partial dependence plots, subpopulation analysis, and individual prediction explanations.
Together, these techniques help explain how a model makes decisions and enable data scientists and stakeholders to understand the factors influencing model predictions.
Powerful what-if analysis allows both technical users and business experts to test different combinations of inputs and review the impact to predicted results. Simulation capabilities even enable teams to systematically uncover and prescribe changes that would lead to the desired business outcomes.
With this visual interface for what-if analysis, business users can build trust in predictive models and apply knowledge of model behavior in practical ways.
Dataiku provides analyses to uncover segments that may be unfairly or differently treated by a model. Interactive subpopulation analysis allows users to compare model results by group, and disparate impact analysis measures whether a sensitive group receives a given outcome at a rate close to that of the advantaged group.
With this information in hand, data scientists can produce models that deliver more responsible and equitable outcomes.
Dataiku generates row-level prediction explanations (ICE and SHAP) to provide additional information for predicted results.
For both batch and realtime scoring, prediction explanations can be returned as part of the response, which fulfills the need to have reason codes in regulated industries and provides additional information for analysis.
Dataiku automatically generates comprehensive documentation for models and the project flow on a one-off or scheduled basis. Customizable templates include all the metadata and visualizations needed to snapshot the project state or model design and results.
With robust auto-documentation, organizations maintain consistent records of AI projects for regulatory compliance, and teams can stop spending countless hours maintaining project docs.
Each Dataiku project has a visual flow that transparently represents the pipeline of data transformations and movement from start to finish. Every design decision and step are captured and displayed so current or future team members can clearly follow and explain the sequence of project logic.
This blog series covers what happens when AI systems make a mistake, and why that matters.
Read the SeriesDiscover the tradeoffs between white-box and black-box models, Dataiku's approach to explainable AI features, and more.
Watch the WebinarIn this ebook, learn how to reduce business risk and infuse transparency into AI initiatives.
Get the E-BookStart Your Dataiku 14-Day Free Trial
or Install the Free Edition of Dataiku