Home // International Journal On Advances in Systems and Measurements, volume 12, numbers 3 and 4, 2019 // View article

Forecasting Transportation Project Frequency using Multivariate Modeling and Lagged Variables

Authors:
Alireza Shoajei
Hashem Izadi Moud
Ian Flood

Keywords: Multivariate Regression; Elastic Net Regularization; Strategic Planning; Project Portfolio Management; Forecasting; Support Vector Machine; Time Series.

Abstract:
Knowledge of the number of upcoming projects and their impact on the company plays a significant role in strategic planning for project-based companies. The current horizon of planning for companies working on public projects are the latest advertised projects for bidding, which in many cases is reported less than a year in advance. This provides a very short-term horizon for strategic project portfolio planning. In this research, a multivariate regression model with elastic net regularization and a support vector machine are used to forecast the Florida Department of Transportation’s (FDOT) number of advertised projects in the future considering economic indices and other environmental factors. Two sets of analyses have been conducted, one with the current values of the independent variables and another one with up to 12 months lag of each independent variable. The results show that, of the predictors considered, the unemployment rate in the construction sector and the Brent oil price are the most significant variables in forecasting FDOT’s future project frequency using current values. Also, it is evident that including lagged values of the independent variables increase the model’s performance.

Pages: 148 to 157

Copyright: Copyright (c) to authors, 2019. Used with permission.

Publication date: December 30, 2019

Published in: journal

ISSN: 1942-261x