The Navy seeks technology based on statistical or computational methods to assist in the continued tracking of training performance and proficiency trends as underlying tactical data changes. OASYS, INC. and the ITCS at UAH proposes to exploit the benefits of modeling the underlying cause-effect structure of Navy data, rather than the data itself. This approach makes the model and analytical methods invariant to changes to in the input distribution, allows for the accurate adjustment of counfounding factors, and enables for the prediction of the outcome of data-driven decisions. The benefits of identifying the cause-effect relationships between variables have been known for some time and are used in many scientific fields where accurate decisions are critical. The method of cause-effect discovery in those fields are often through randomized experimental trials, but the development of causal discovery methods that infer causal relationships from uncontrolled data (non-experimental) is an important and growing area of research that shows great promise in data analytics.
Benefit: Big data analytics and machine learning are now ubiquitous in most U.S. industries. These analytics methods hold considerable promise in all areas in which Team OASYS is involved: armed services, missile defense, homeland security, commercial IT network infrastructure, and pharmaceutical research. However, traditional data modeling techniques will continue to fall short of the mark because, by their construction, they cannot infer cause-effect relationships that are the difference between simple correlation and actionable analysis. Bringing the proposed cause-effect inference methodology to maturity in the proposed STTR research will open all of these commercialization avenues, of which the defense market in Huntsville alone has volume in excess of $2 billion annually.
Keywords: Big data analysis, Big data analysis, Causal discovery, causal inference, data trends, Machine Learning, statistical analysis, data predictions