DTIC ADA395566: The Application of Information Mining Technology to the Army Injury and Health Outcomes Database

📝 Description

In the first half of the grant period we faced many challenges, principally administrative in nature, and mostly related primarily to delays in the procurement of equipment, software, and the awarding of contracts. A significant amount of training was accomplished, and while the learning curve was steep, several individuals received sufficient basic instruction to at least initiate the various processes needed to successfully accomplish the objectives of the proposal. In addition, a special on-site block of instruction was arranged with SAS Institute. This instruction, known as a pilot was divided into two parts; a Warehouse Administrator (WA) pilot, and an Enterprise Miner (EM) pilot. Each pilot consisted of 5 full days of training and hands-on assistance. The WA pilot was designed to build the TAIHOD data warehouse and to provide instruction to the designated TAIHOD Warehouse Administrator (Ms. Yore). The EM pilot, was designed to walk several members of the TAIHOD staff through the process of a data mining exercise using data from the warehouse in the TAIHOD environment (Ms. Yore, Mr. Schneider, and LTC Amoroso). To get the full benefit of the mining pilot, it was necessary to select a topic suitable for demonstrating the various idiosyncrasies and capabilities of the software. We chose a practical, though arguably esoteric, exercise to examine data quality. Specifically, we attempted to identify patterns of gender misclassification across the various components of the database. This exercise had the potential utility of allowing us to find patterns of error in the database that, if corrected, could greatly improve the precision of certain analyses where there is ambiguity in the process of matching records. It also involved the use of multiple components of the database, including text (names). Several documents attached to this report provide substantial detail on this very much still-in-progress effort.