Predictive Analytics

Forecasting or predicting future activities is a long standing challenge. Analysts require new tools, processes, and methodologies to help improve capabilities in this area.  Our efforts will focus on developing novel tools, techniques and methods for characterizing and reducing uncertainty. Activities under this thrust aim to provide decision makers with timely and accurate forecasts of significant events. Predictive analytics are desired for application across a multitude of topics.

Current analytical tradecraft is inherently retrospective and reactionary. Predictive Analytics will enable development of technologies to generate timely forecasts for well-defined events and their characteristics (i.e., who, what, when, where, and how.) This will be accomplished through a rigorous, open and ongoing test and evaluation process. Forecasts will be communicated in the context of intelligence community interests, and will rely on both data at rest and streaming data.

Predictive analytics must be verifiable, measurable, and repeatable, but results must also be simple and streamlined enough for analysts to understand; work with visualization and situational awareness teams will be needed. Predictive analytics will only be as good as the data they’re relying on, so close collaboration with analysts generating the data used for input will be vital. LAS is working to develop new approaches for predictive data analysis, prove the feasibility of accurately anticipating events based on statistical analysis of data, and identify significant policy and tradecraft changes that may be required to support predictive analytics.