Data Mining
Aegis brings decades of experience in identifying patterns and relationships in data. We use a range of statistical techniques to discover previously unknown and potentially useful information. Through this knowledge discovery process we create models to predict future behavior. We have applied data mining to:
- Develop system to identify and estimate royalty underpayment on Federal oil leases for statistical billing.
- Analyze caseload processing and develop models predicting growing backlogs resulting from a faulty $200M state benefit-processing system.
- Conduct a study of minority voting patterns to reveal patterns of racial bloc voting resulting in major redistricting.
- Develop a novel method for revealing socioeconomic and age bias in jury selection.
- Analyze detailed records for over 150,000 students to reveal patterns relating race, poverty and educational achievement.
- Analyze property management records for patterns of treatment of persons with disabilities.
- Discover characteristics of households for a survey of immigrant families that removed 97% ineligible households, by data mining commercially available marketing databases.
- Analyze the relationship between stamp duty rates and real estate sales.
- Analyze election records for evidence of mishandling or fraud in elections.
- Analyze employment records for patterns in employment decisions.
- Construct a model of jury selection process revealing systemic bias.
- Develop algorithms applying probability theory to improve performance of genetic programming application.
