Framework for Integrating an Artificial Neural Network and a Genetic Algorithm to Develop a Predictive Model for Construction Labor Productivity

摘要

Construction labor productivity (CLP) is one of the most important factors in the construction industry, as it has a direct effect on a company's efficiency and profitability. The accurate prediction of CLP is essential for effective decision-making prior to project execution, and continuous tracking and improvement of productivity over a project life cycle is necessary for its success. The objective of this paper is to develop a framework to help construction organizations predict and measure construction productivity, leading to improved project performance in terms of cost, time, and quality. CLP is affected by numerous factors, including the high-dimensional factors that result from a large number of model input variables and which often impose a high computational cost and the risk of overfitting of data. Therefore, it is necessary to use feature selection methods to reduce the dimensionality of CLP data. This paper proposes a framework that integrates an artificial neural network (ANN) and a genetic algorithm (GA) for feature selection. The proposed framework is used to develop a predictive model for CLP using features selected because they provide the best prediction of CLP. The ability of GAs to generate an optimal feature subset in combination with the superior accuracy of ANNs is a unique advancement that this framework offers for improving the prediction of labor productivity. The developed model can predict productivity and specify which factors are most predictive of CLP. The contributions of this paper are (1) the development of a framework that uses an integrated ANN and GA as a wrapper method for selecting the features with the most influence on CLP and (2) the development of an improved predictive model that can be used to both predict and measure CLP.