The availability of large unstructured datasets accelerated via autonomous data collection and warehousing, makes the use of advanced analytical skill-sets a critical necessity. Data extraction, mining, clustering, modeling, and visualization are important processes for defensibly estimating the lifecycle costs of Federal Government acquisition programs.

The incorporation of data science methodologies can yield meaningful improvements to cost estimating practices by promoting effective data governance, streamlining data collection and normalization, and providing increased opportunities for exploratory analysis through advanced workflows otherwise thought to be unfeasible.

Adherence to traditional paradigms of manual data collection and processing prevent cost analysts from leveraging emerging data science methodologies and programming languages, which in turn, inhibits accessibility to complex and diverse datasets from dynamic repositories. Traditional methodologies – such as the assignment of uncertainty/risk parameters – can result in subjective inputs and unsubstantiated results; whereas statistical models derived from computational algorithms can objectively capture and process vast amounts of raw data for defensible cost estimation and predictive analysis.

By recommending the adoption of a data science paradigm, this presentation will evaluate the software and skill-sets required for advanced computational modeling and analysis. It will compare formal and unstructured training formats to develop workforce aptitudes, and will discuss labor force ratios of data architects, engineers, scientists, and analysts as collective specialists in structured data processing, management, and implementation.