A neural network application to subcontractor rating in construction firms

Fuzzy Multi-Criteria Decision Making Model for Subcontractor Selection in International Construction Projects

“…They have presented a model for analyzing the subcontractor's risk elements in construction refurbishment projects and described a prototype knowledge-based expert system. Albino and Garavelli (1998) used neural networks for the selection of subcontractors. However, these studies have some drawbacks which can be listed in the following items:…”

“…- Ip et al (2004), Tserng and Lin (2002), Okoroh and Torrance (1999), and Albino and Garavelli (1998) considered very few selection criteria and formed one-tier selection procedure, which made the selection process superficial, Technological and Economic Development of Economy, 2016, 22(2): 210-234 - Ip et al (2004) and Tserng and Lin (2002) did not take into account qualitative criteria, which was obviously in contradiction with the characteristics of a real life problem, - Ng and Luu (2008), Okoroh and Torrance (1999), and Albino and Garavelli (1998) depended solely on past data that should be collected and trained to solve the problem, which produced a time-consuming model in the short-and mid-term, - Mbachu (2008) and Ip et al (2004) did not employ any automated system, which made the calculation procedure an effort-consuming process, - Okoroh and Torrance (1999) has project-specific criteria, which did not allow the model to be easily used for all kinds of projects, and - Arslan et al (2008) evaluated criteria that were scored on a 1 to 10 scale, which was, in fact, not adequately suitable for decision makers because of the fact that human perception and judgment cannot be quantified precisely and that decision makers intuitively feel more comfortable providing their judgments in verbal terms (rather than numerically), which, due to subjectivity, leads to ambiguity in human decision making (Poyhonen et al 1997).…”

A neural network approach for early cost estimation of structural systems of buildings

“…The architecture of the neural network model. [11,20,23,24]. The neural network model is empirically, rather than theoretically derived.…”

“…An important issue to be resolved when applying neural networks to a problem is to determine which training procedure to adopt [24]. There are many other alternative paradigms to choose from.…”

“…; nÞ, and EðiÞ is the target output, i.e., the estimated building cost. The mean square error is a good overall measure of whether a training run was successful [16,24,30]. It was used for evaluating the performance of the model during the training process.…”