Bidding Model Incorporating Bid Position for Determining Overhead-cum-markup Rate

Modeling bidding decisions and bid markup size for construction projects: A fuzzy approach

“…ANFIS has been shown to perform better with historical quantitative data sets from past similar construction projects [14], which is a major limitation for many contractors as construction projects tend to lack records of specific markup factors. Chao et al [6], on the other hand, propose a very similar FIS model to the present study, but consider only two factors to develop the fuzzy rules. In the absence of quantitative data sets from past projects, the proposed Mamdani FIS model can work better in capitalizing on experts' judgments.…”

“…Most markup studies end with finding a potential and critical factor associated with the d2b and estimating the markup. Some studies propose fuzzy-based markup prediction models [5,6,14], with the Fuzzy Neural Network, for instance, demonstrated on 270 hypothetical bidding data sets, where only three factors were considered for markup prediction [5]. ANFIS has been shown to perform better with historical quantitative data sets from past similar construction projects [14], which is a major limitation for many contractors as construction projects tend to lack records of specific markup factors.…”

“…The contractor is responsible for the bid preparation and must consider all costs and risks involved. Traditionally, depending on their experiences and intuitions, each contractor adds some percentage of the estimated project cost as a markup, which defrays overheads, contingencies, and targeted profit [4][5][6]. Although traditional experience-based markup selection may work well for some contractors, a comprehensive understanding of the project's context, risks, and uncertainties involved in its execution phases helps contractors find an optimal markup.…”

“…On the other hand, fuzzy-based models can manage expert-judgment-based data sets and have been treated as of potential value in previous studies. Accordingly, Arabpour and Moselhi [14], Al-Humaidi [15], and Chao and Liaw [6] have proposed three different fuzzy-based markup estimation models. However, Arabpour and Moselhi's [14] fuzzy-ANN model is data-intensive, and depends on a significant number of quantitative data sets for training and testing purposes; Al-Humaidi's [15] model can manage only a maximum of 10 inputs for markup estimation; while Chao and Liaw's [9] model considers only two factors without finding all the critical factors involved.…”