A Hybrid Neural Network for Predicting Construction Labour Productivity

Abstract

This paper presents a novel Computational Intelligence (CI) approach to model construction labour productivity. A hybrid neural network combining the General Regression Neural Network (GRNN), Fuzzy Logic (FL) and Genetic Algorithms (GA) are used to identify and quantify factors affecting construction labour productivity and to predict performance. The essential features of the network are described in detail. Its reasoning capability, predictive behaviors, and advantages are discussed. The use of the network is demonstrated for an example project.