Big Data Analysis of Contractor Performance for Services Acquisition in DoD: A Proof of Concept

Abstract

This paper examines the use of Big Data analytic techniques to explore and analyze large datasets that are used to capture information about DoD services acquisitions. It describes the burgeoning field of Big Data analytics, how it is used in the private sector, and how it could potentially be used in acquisition research. It tests the application of Big Data analytic techniques by applying them to a dataset of CPARS ratings of acquired services, and it creates predictive models that explore the causes of failed services contracts using three analytic techniques: logistic regression, decision tree analysis, and neural networks. The report concludes that four variables exhibit the largest impact on the success/failure rates of services contracts: type of contract; awarded dollar value; workload per filled billets; % of 1102 billets filled by contracting office.