Modeling and Prediction of Asymmetric Threat Learning Processes

CCRIs Phase I research efforts on Modeling and Prediction of Asymmetric Threat Learning Processes have shown the potential for learning models to improve predictive accuracy. The methodology developed proved to be robust and provides improvements in prediction, even when parameter estimation for the structural equation models was not significant. The Phase I results suggest that the development of a prototype tool based on the methods CCRI developed is necessary to automate information selection and build a database to store large volumes of data, model parameters and results. The proposed tool will also enable formal testing of the model for robustness using alternative datasets and scenarios. Several questions arose from the Phase I work and provide a basis for additional Phase II technical objectives. These include measuring and evaluating the effectiveness of model components and other factors that might influence enemy attacks.