A Predictive Logistic Regression Model of World Conflict Using Open Source Data
Deteriorating and failing federal facilities represent a cost to leaders and organizations as they attempt to manage and maintain these assets. Currently the Air Force employs the BUILDERTM Sustainment Management System to predict the reliability of building components. At different system levels, however, the probabilities of failure are not predicted. The purpose of this research is to provide probabilistic models which predict the probability of failure at the system level of a building's infrastructure hierarchy. This research investigated the plumbing, HVAC, fire protection, and electrical systems. Probabilistic models were created for these systems by using fault trees with fuzzy logic on the basis of risk by weighting the probabilities of failure by the consequences of failure. This thesis then validated each of the models using real-world Air Force work order data. Through contingency analysis, it was found that the current BUILDERTM condition index model possessed no predictive ability due to the resulting p-value of 1.00; the probabilistic models possessed much more predictive ability with a resulting p-value of 0.12. The probabilistic models are statistically shown to be a significant improvement over the current condition index model; these models lead to improved decision making for infrastructure assets.
- Paperback | 102 pages
- 215.9 x 279.4 x 5.84mm | 326.58g
- 02 Jun 2015
- United States
- black & white illustrations