The DARPA Data Driven Discovery of Models (D3M) program automates methods in data science to enable domain experts to incorporate their knowledge into the modeling process and create meaningful and valid predictive models of real, complex processes without the need for expert data scientists.
Challenge: Current analytic workflows involve different experts with partial knowledge using different tools and many manual steps.
Solution: D3M reinvents the data science tools and processes to achieve effective iteration in model building. D3M delivers single integrated platforms that are fully accessible by domain experts. With automation that handles real world problems and delivers accurate answers in minutes rather than months.
D3M automated modeling and decision tools allows domain experts to quickly iterate between descriptive ("what happened"), predictive ("what might happen"), and prescriptive analytics ("what should we do"). D3M platform tools are built on top of powerful, open, and extensible AutoML engines with the ability for data scientists to build custom applications or analytics to support domain experts for unique requirements or specialized pipelines. The D3M human-guided machine learning platforms provide automation that ensures statistically valid models and results that are explainable and trustworthy. The collaborative workflows support a multi-disciplinary approach to modeling and decision making.