Innovative Data Anomaly Detection and Transformation for Analysis Applications

NAVAIR"s Warfare Analysis and Integration Department (AIR 4.10) has three labor intensive, error prone, data management operations that Sonalysts"desktop, Windows application will automate with a generic toolkit to greatly reduce the human effort. The data provided by the fleet tends to be error prone, but modeling, simulation, and analysis require valid data. The data contains both quantitative and qualitative elements which are interdependent. Additionally, these processes must be run on hypothetical datasets to evaluate proposed equipment changes, where the potential benefits are loosely defined. Once the data is cleaned, the results need to be formatted to support downstream analysis tools. The codes, rules, and formats evolve over time. Sonalysts proposes a set of user-configurable, rule-set generators for creating and maintaining instructions for cleaning the data, synthesizing hypothetical data, and formatting output datasets. These rule sets are used by the code, but are not part of it. Thus, modifications do not require program source code changes or recompiling. Qualitative data analysis techniques will be used to perform weighted matching of text elements to automate selection of dependent code values.