Forecasting Construction Cost Index through Artificial Intelligence

• Published in: Societies (Basel, Switzerland) Switzerland), 2023-10, Vol.13 (10), p.219
• Main Authors: Aslam, Bilal, Maqsoom, Ahsen, Inam, Hina, Basharat, Mubeen ul, Ullah, Fahim
• Format: Article
• Language: English
• Subjects: Accuracy; Algorithms; Analysis; Artificial intelligence; artificial neural network (ANN); Artificial neural networks; Budgets; Building materials; Company pricing policy; Construction; construction cost index (CCI); Construction costs; Construction industry; Construction materials; Contractors; Cost accounting; Cost estimates; cost estimation; Costs; Developing countries; Documentation; Econometrics; Forecasting; Forecasts and trends; Gravel; Highway construction; Indexes; Inflation; Inflation (Finance); LDCs; Methods; Neural networks; Pakistan; Price indexes; Prices; Prices and rates; Process costing; Regression analysis; Regression models; Reinforced concrete; Statistical analysis; Support vector machines; Technology application; Time series; Urban planning; Values; Variables
• ISSN: 2075-4698; 2075-4698
• DOI: 10.3390/soc13100219

This study presents a novel approach for forecasting the construction cost index (CCI) of building materials in developing countries. Such estimations are challenging due to the need for a longer time, the influence of inflation, and fluctuating project prices in developing countries. This study used three techniques—a modified Artificial Neural Network (ANN), time series, and linear regression—to predict and forecast the local building material CCI in Pakistan. The predicted CCI is based on materials, including bricks, steel, cement, sand, and gravel. In addition, the swish activation function was introduced to increase the accuracy of the associated algorithms. The results suggest that the ANN model has superior prediction results, with the lowest Mean Error (ME), Mean Absolute Error (MAE), and Theil’s U statistic (U-Stat) values of 0.04, 28.3, and 0.62, respectively. The time series and regression models have ME values of 0.22 and 0.3, MAE values of 30.07 and 28.3, and U-Stat values of 0.65 and 0.64, respectively. The proposed models can assist contractors, project managers, and owners through an accurately estimated cost index. Such accurate CCIs help correctly estimate project budgets based on building material prices to mitigate project risks, delays, and failures.