Developing forecasting models for PFI data in Sabah Region

• Main Authors: Md Ghani, Nor Azura, Ramli, Norazan Mohamed, Bin Ahmad Kamaruddin, Saadi
• Format: Conference Proceeding
• Language: English
• Subjects: artificial neural network; Biological system modeling; construction; Estimation; forecast; Forecasting; Materials; Neural networks; Predictive models; price indices; Private Financial Initiative; Steel
• DOI: 10.1109/IC3e.2013.6735971

One prominent phase of which due attention is required in the Malaysian Private Financial Initiative agenda is value for money, under which aspects like optimum efficiency and effectiveness of each expense have been well accomplished. In this paper, the main objective lies in approximating unitary charges or materials' price indices in each Malaysian territory. Here, the goal is to find out the best forecasting method that can best be adopted to calculate the unitary charges price indices of the construction industry in the Malaysian Sabah Region. The data of the unitary charges indices were monthly data from year 2005 to 2011 concerning a range of construction material price indices in Sabah. The data comprise the price indices of aggregate, sand, steel reinforcement, ready mix concrete, bricks and partition, roof material, floor and wall finishes, ceiling, plumbing materials, sanitary fittings, paint, glass, steel and metal sections, timber and plywood. The concluding part of this paper suggests that the backpropagation neural network with linear transfer function was proven to establish results that are the most accurate and dependable for estimating unitary charges price indices in Sabah based on the Root Mean Squared Errors, where both the estimation and evaluation set values were roughly zero and highly significant at p