Assessed the USPTO's transition from the USPC to CPC code system, evaluated the accuracy of their statistical mapping system, and evaluated how well both patent classification systems discriminate between technology areas. Used SAS to parse XML data and Python's Natural Language Took Kit (NLTK) for analysis.
This projection of US military ground vehicle procurement is synthesized from budget data, military program office communications, congressional markups and testimony, and contract awards. Scenarios, as well as the budget as requested, are provided annually.
Worked with contracting, program management, logistics, and engineering offices to develop an data warehouse ER diagram to support soldier equipment. Joined all data sources and produced an initial report to illustrate value and intended output. Mocked this up the old fashioned way: Microsoft Excel and Microsoft Access.
Evaluated the innovation strategy of leading aerospace/defense contractors by benchmarking their patent portfolio against the US Air Force R&D budget. Used topic modeling, clustering, and text/data mining to develop metrics that indicate patent quality. Utilized SAS, MALLET, R and some JAVA help from friends.
Co-authored a paper with US Army Research Lab staff that forecasts when the US may become involved in the next war or contingency operation by statistically analyzing US war activities and DOD funding from 1805 to 2017. Most beltway think tanks only look back to 1947; we worked harder and created a database of DOD funding that reaches back to 1805.
We quantified the impact on retail sales of three crises that caused changes in consumer sentiment and purchasing behavior: the US 911 attacks in 2001, the US Severe Acute Respiratory Syndrome (SARS) scare of 2003, and the Hong Kong SARS outbreak of 2003. We then forecasted the effect of a potential ebola outbreak or scare on US retail sales.
This report is a demand based forecast of the global body armor market required to support the military, the paramilitary, and law enforcement agencies within 52 countries. We used an index driven methodology and a comprehensive and credible set of international references and data sources to develop this forecast.
Analysis of 16 million inpatient and emergency room (ER) visit records to determine if certain diagnosis codes in the ER were early/likely indicators of an eventual procedure that required inpatient treatment. MYSQL drove this analysis.
Used predictive analytics to identify how the organization's activities supported the veteran's outcome. Determine which activities and veteran characteristics contribute most to successful outcome. Developed a recommender system to match veterans with potential employers. Used SAS, MYSQL, and a little PHP/HTML.
Vector Analytics leverages data mining and data science techniques to provide strategic insight to organizations in the materials, defense, and aerospace industries.