In Turkey, for the preliminary construction cost estimation, a notice, which is updated and published annu­ally by Turkish Ministry of the Environment and Urbanism, known as “unit area cost method” (UACM) is generally employed. However, it’s known that the costs obtained through this method in which only construction area is taken into consideration have significant differences from actual costs. The aim of this study is to compare the cost estimations obtained through “multi layer perceptron” (MLP) and “radial basis function” (RBF), which are commonly used artificial neural network (ANN) methods. The results of MLP and RBF were also compared with the results of UACM and the validity of UACM was interpreted. Dataobtained from 232 public construction projects, which completed between 2003 and 2011 in different regions of Turkey, were reviewed. Consequently, estimated costs obtained from RBF were found to be higher than the actual costs with a 0.28% variance, while the estimated costs obtained from MLP were higher than actual values with a 1.11% variance. The approximate costs obtained from UACM are higher than actual costs with a 28.73% variance. It was found that both ANN methods were showed better performance than the UACM but RBF was superior to MLP.

This paper investigates the adequacy of some nontraditional approaches to produce realistic forecasts of the final costs of building projects and compare them with forecasts produced by three traditional methods, the unit area costs (UAC), client detailed costs (UPA), and contract sums (CS). As a case study, data of the actual final costs and forecasts produced using the three traditional methods for 420 finished public building projects carried out in Turkey were collected. Based on 75% of the collected data (i.e., 316 projects), three cost models based on UAC, UPA, and CS were established. The remaining 25% of the data (104 projects) were used as control data for testing the accuracies of forecasts produced by the different approaches. Cost forecasts using five nontraditional approaches, namely multilayer perceptron (MLP), radial basis function (RBF), grid partitioning algorithm (GPA), reference class forecasting (RCF), and regression analysis (RA) were then produced and compared with the actual...

Accurate estimation of construction costs is of foremost importance in construction management processes. Considering the changes and unexpected situations, cost estimations should be revised during the construction process. This study investigates the predictability of earned value management (EVM)-based approaches using machine learning (ML) methods. A total of 2318 data points via 19 EVM-based cost estimation methods were created and six ML methods were used for the analyses. The planned and actual project data of the rough construction activities of a housing project completed in Türkiye were used. The ML methods considered consisted of adaptive neuro-fuzzy inference systems (ANFISs), artificial neural networks (ANNs), Gaussian process regression (GPR), long-short-term memory (LSTM), M5 model trees (M5TREEs), and support vector machines (SVMs). The created models were compared using performance criteria such as mean absolute percentage error (MAPE), relative root means square error (RRMSE), coefficient of determination (R²), Nash–Sutcliffe efficiency coefficient (NSE), and overall index of model performance (OI). Moreover, radar charts, trend graphs, Taylor diagrams, violin plots, and error boxplots were used to evaluate the performance of the estimation models. The results revealed that the classical ANN model outperforms EVM-based cost methods that utilize current ML methods.