EACE Journal — Vol. 10, No. 1
COST ESTIMATION OF HIGHWAY PROJECTS USING ARTIFICIAL NEURAL NETWORK COUPLED WITH BOOTSRAP & MONTE CARLO SIMULATION
COST ESTIMATION OF HIGHWAY PROJECTS USING ARTIFICIAL NEURAL NETWORK COUPLED WITH BOOTSRAP & MONTE CARLO SIMULATIO
Abstract
Some applications and mathematical models are limited in simulating the complex behavior of problems. However, various studies have shown how Artificial Neural Network (ANN) has significant benefits and makes it a practical tool for solving complex problems in the field of engineering and construction. Although ANN is a powerful computing system, it has its own limitations. For example, ANN is frequently criticized as being “black box”. In addition, ANN does not sufficiently incorporate the effect of the learning curve and the time factor in predicting ongoing operations when objective data like historical data and/or information are at hand. This study aims at coupling ANN with bootstrap (BS) resampling to enhance its performance and prediction capability by taking highway project cost data as a case study. The accrued data were collected from Ethiopian Road Authority’s (ERA) Procurement Compilation Report (PCR) and Engineering Report, which contain contract cost data and project related parameters. Monte Carlo Simulation is also utilized to assess the capability of ANN in detecting meaningful differences and thereby increase its ability to detect sensitivity in changes when it really exists. A range of parameters were set at different possible scenarios and the outputs were simulated. The study identified that neural network ap proach when coupled with bootstrap resampling and simulation provided better prediction. The enhanced models showed a far more consistent performance and produced smaller Mean Absolute Percentage Error (MAPE) of 0.18 in comparison with the standard models (0.20). This approach is crucial for addressing the several parameters that can influence construction project costs and in turn develop close-to-accurate cost models in which the developed model can be applicable in estimating costs for future projects. Key Words: Artificial Neural Network (ANN), Bootstrap, Monte Carlo simulation, Cost Estimation, Road Project
Read Full Article (PDF)