DataRobot and Palantir share a philosophy that AI value is delivered by uniting an organization’s data scientists, decision-makers, and everyday employees in an environment of collaboration on AI/machine learning-powered operations.

Today, the two organizations are excited to announce that they have partnered together to tackle one of the most complex challenges that retailers face: demand forecasting. 

While many retailers already rely heavily on machine learning and AI to forecast demand, the COVID-19 pandemic has shown us the vast limitations of existing models and training data that have been built on the premise of predictable historic patterns and behaviors. None of us could have foreseen the immense changes in the way we shop nor the increasing demand for certain products that the pandemic brought about. But the reality is that the way retailers forecasted demand pre-COVID was problematic; the pandemic simply brought into sharper focus the need for smarter and more real-world solutions to an age-old problem.

Live visibility into supply and demand requires bringing together datasets from different source systems, with different permissions and update cycles. The complexity of bringing together these assets to provide a real-time and comprehensive view of the retail ecosystem is insurmountable for most organizations. And without a holistic view and a common language between teams, data scientists, subject matter experts and business decision-makers are left with blind spots when it comes to critical decision-making.

DataRobot and Palantir have teamed up to take these issues head on with their Demand Forecasting solution. This newly developed, custom framework brings together the best of Palantir Foundry and DataRobot Model Development and Model Deployment capabilities to manage the most complex of retail use cases from a single platform.

With Demand Forecasting, retailers can:

Create and deploy machine learning models easily with an end-to-end integrated solution.

Users can easily create end-to-end AI workflows, including data prep and automated model development, all the way to model deployment and forecasting.

Workflows can be operationalized quickly and effectively, moving beyond dashboards and business intelligence into real-world applications driving decisions. And with all machine learning assets in one place, model monitoring and ongoing management is vastly simplified.

Scale your approach rapidly with automation and smart integration.

Once your data prep is finalized, you can pipe the data directly from Palantir Foundry to a unique DataRobot project. Once the project is configured, DataRobot’s Autopilot feature uses built-in guardrails and data science best practices to ensure that users don’t miss critical steps in the modeling process.

After the optimal model framework is identified, it is deployed to a dedicated prediction environment and is ready to serve predictions back down to Palantir Foundry using fresh data. As new data from Palantir Foundry is scored, you can monitor and access the model’s performance by tracking data drift and accuracy.

Get to value faster by uniting multiple users around a single view of the ecosystem.

A digital twin of the organization is created with a semantic data layer that weaves together critical data and analytics assets in a readily understandable format.

With a single and accessible source of truth, non-technical subject matter experts can easily view and understand the forecasts to make business-critical decisions faster and with more confidence.

Together, DataRobot and Palantir Foundry can transform the demand forecasting process for retail organizations, creating a more intelligent and more impactful solution for all.

To find out more about the Demand Forecasting solution, visit https://www.datarobot.com/palantir-forecasting/