A Comparison of Projection Methods in the Forecasting of Overhead Costs for Seven Government Aerospace Contractors.

Abstract

This thesis compares three types of forecasting models developed to predict overhead costs for seven government aerospace contractors. The methodologies utilized to develop the models include generalized least squares, univariate Box-Jenkins, and multivariate Box-Jenkins procedures. The results of those models are compared using three measures of effectiveness: correlation coefficient between actual and predicted values, root mean squared error divided by the mean of the actuals, and mean absolute percentage error (in percent). As was expected, the univariate Box-Jenkins method produced short term forecasts which were superior to those of the least squares regression models. However, the regression forecasts were highly accurate and were considerably less expensive to obtain. Only one multivariate Box-Jenkins model could be developed . The results of this model were marginally superior to the related regression model and significantly inferior to the univariate Box-Jenkins model for the same contractor. Keywords: Thesis; Autocorrelation; Transfer functions; Charts.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Sep 01, 1987
Accession Number
ADA187572

Entities

People

  • Christopher P. Schnedar

Organizations

  • Naval Postgraduate School

Tags

Communities of Interest

  • Human Systems
  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Aerospace Industry
  • Boxes
  • Coefficients
  • Computers
  • Contractors
  • Data Science
  • Data Sets
  • Differential Equations
  • Econometrics
  • Equations
  • Governments
  • Information Science
  • Regression Analysis
  • Standards
  • Statistics
  • Transfer Functions
  • White Noise

Readers

  • Life Cycle Cost Analysis
  • Regression Analysis.

Technology Areas

  • Space