A Cascaded Neural Network on Labour Cost Forecasting in Chongqing’s Construction Industry

摘要：

The ability to predict labour cost fluctuation can result in competitive advantage in the modern construction market and avoid under-or overestimation. In this paper, a new forecast strategy is proposed for monthly prediction of labour cost and Chongqing抯 construction industry is taken as a case study. Every month..Chongqing Engineering Cost Information(CECI) publishes the labour cost of general worker ,which is used as the target variable in this paper. Since labour cost has a nonlinear and time dependent behavior, considering the influence of regional economy..through Literature Study Gross Domestic Product, Rate of Registered Unemployment, General Consumer Price Index, Average Wages and Overall Labour Productivity of Chongqing are regarded as the candidate variables to forecast the labour cost. This forecast strategy is composed of a preprocessor and a Cascaded Neural Network (CNN). Preprocessor selects the input features of the CNN according to MRMR (Maximum Relevance Minimum Redundancy) principal. The CNN is composed of two neural networks in a cascaded structure. Finally, the proposed cost forecast strategy is compared with the most common technique-simple exponential regression in the area and we can draw a conclusion that it is a more accurate and improved system for labour cost forecasting.

关键词： labour cost neural network LM BFGS.

作者: Shirong LI Lanyue LIU

作者单位: Faculty of Construction Management and Real Estate,Chongqing University, Chongqing, China 400045 Faculty of Construction Management and Real Estate, Chongqing University, Chongqing, China 400045

会议类型: 国际会议

会议名称: 2011 International Conference on Construction & Real Estate Management(2011建设与房地产管理国际会议)

会议地点: 广州

会议语种:英文

页码: 497-501

在线出版日期: 2011-11-19（万方平台首次上网日期，不代表论文的发表时间）