Trefor P. Williams

Predicting final cost for competitively bid construction projects using regression models

Abstract This paper studies the relationship between the low bid and completed cost for competitively bid highway projects. Data are analyzed from several agencies managing highway and dredging projects. A natural log transformation of the low bid and final cost was found to produce regression models for each agency that had high R and R 2 values. These models can predict completed project cost using only the project low bid as input. A separate regression model was determined for each agency. Based on the form of the calculated models it appears that highway agencies construct projects where the final cost increases as an increasing percentage of the low bid price as the project magnitude increases. The dredging projects appear to follow different mechanisms of cost increase than the highway projects and were not predicted as well by the regression models. The regression models can also be used for budgeting purposes by submitting the sum of the low bids for a group of projects as input. The regression models output a prediction of the cost of the group of projects that was found to be highly accurate.

Predicting completed project cost using bidding data

Neural network and regression models have been developed to predict the completed cost of competitively bid highway projects constructed by the New Jersey Department of Transportation. Bid information was studied for inclusion as inputs to the models. Data studied included the low bid, median bid, standard devi9 ation of the bids, expected project duration and the number of bids. A natural log transformation of the data was found to improve the linear relationship between the low bid and completed cost. The stepwise regression procedure was applied, and yielded the best performing predictive model. This regression model used only the natural log of the low bid as independent variable to predict the natural log of the completed cost. Radial basis neural networks were also constructed to predict the final cost. The best performing regres4 sion model produced superior predictions to the best performing neural network model. Hybrid models that used a regression model prediction as an input to a neural network were also studied and were found to also produce reasonable predictions. The calculated models produced good predictions of the completed project cost, but were found to be deficient in predicting very large cost increases. Simple models using the natural log of the low bid as input produced the best results. From the analysis it may be concluded that additional information about the variability of the bids submitted does not provide useful information for predicting the final project outcome.

Differing perspectives on collaboration in construction

Purpose – The purpose of this research is to identify the key aspects present in collaborative projects with the objective of producing a clear definition for collaboration within the UK construction industry. Firstly, the research provided a summary of the different forms of “working together” that have become more prevalent since Lathams and Egans work. Partnering was seen as the ultimate form of collaboration, but due to the recent economic crisis, it has enjoyed diminishing support. Collaboration was perceived as the new way forward. However, the literature on the subject often used the term “collaboration” interchangeably with partnering, alliances, joint ventures, and networks. Therefore, the aim of this research is to identify what the meaning of collaboration is currently.Design/methodology/approach – Primary research was carried out in order to provide a clearer picture of what collaboration is. Both qualitative and quantitative data was collected through a triangulation of questionnaires and i...

Bidding ratios to predict highway project costs

Purpose – Ratios were constructed using bidding data for highway construction projects in Texas to study whether there are useful patterns in project bids that are indicators of the project completion cost. The use of the ratios to improve predictions of completed project cost was studied.Design/methodology/approach – Ratios were calculated relating the second lowest bid, mean bid, and maximum bid to the low bid for the highway construction projects. Regression and neural network models were developed to predict the completed cost of the highway projects using bidding data. Models including the bidding ratios, low bid, second lowest bid, mean bid and maximum bid were developed. Natural log transformations were applied to the data to improve model performance.Findings – Analysis of the bidding ratios indicates some relationship between high values of the bidding ratios and final project costs that deviate significantly from the low bid amount. Addition of the ratios to neural network and regression models ...

Using neural networks to predict component inspection requirements for aging aircraft

Abstract Currently under development by the Federal Aviation Administration (FAA), the Safety Performance Analysis System (SPAS) will contain indicators of aircraft safety performance that can identify potential problem areas for inspectors. The Service Difficulty Reporting (SDR) system is one data source for SPAS and contains data related to the identification of abnormal, potentially unsafe conditions in aircraft or aircraft components/equipment. A higher expected number of SDRs suggests a greater possibility of a maintenance problem and may be used to alert Aviation Safety Inspectors (ASIs) of the need for preemptive safety or repair actions. The preliminary SDR performance indicator in SPAS is not well defined and is too general to be of practical value. In this study, an artificial neural network model is created to predict the number of SDRs that could be expected by part location using sample data from the SDR database that have been merged with aircraft utilization data. The predictions from the neural network models are then compared with results from multiple regression models. The methodological comparison suggests that artificial neural networks offer a promising technology in predicting component inspection requirements for aging aircraft.

Is incentivisation significant in ensuring successful partnered projects

Purpose – This research aimed to test the hypothesis “The use of incentivisation with a gain/pain share of about 15 per cent is a precursor to the achievement of successful infrastructure partnering projects in South Wales”. This hypothesis arose from Egans speech in 2008 discussing the success of partnering.Design/methodology/approach – Two infrastructure projects in South Wales were chosen for the study. This research demonstrates that partnering is not suitable for all projects. Incentivisation places a focus on cost that can have a detrimental effect on the other aspects that exist within the oft quoted triangle of time, cost and quality.Findings – Neither of the two case projects can be judged a success from the perspective of both parties. What represents success to one client would not equal success to the other. Overall it must be concluded that the hypothesis was not proven. Egans view appears to be too simplistic to apply in all situations and is not always the key to success as he suggests.Or...

Comparison of regression and neural network models for prediction of inspection profiles for aging aircraft

Currently under phase 2 development by the Federal Aviation Administration (FAA), the Safety Performance Analysis System (SPAS) contains ‘alert’ indicators of aircraft safety performance that can signal potential problem areas for inspectors. The Service Difficulty Reporting (SDR) system is one component of SPAS and contains data related to the identification of abnormal, potentially unsafe conditions in aircraft and/or aircraft components/equipment. SPAS contains performance indicators to assist safety inspectors in diagnosing an airlines safety ‘profile’ compared with others in the same peer class. This paper details the development of SDR prediction models for the DC-9 aircraft by analyzing sample data from the SDR database that have been merged with aircraft utilization data. Both multiple regression and neural networks are used to create prediction models for the overall number of SDRs and for SDR cracking and corrosion cases. These prediction models establish a range for the number of SDRs outside which safety advisory warnings would be issued. It appears that a data ‘grouping’ strategy to create aircraft ‘profiles’ is very effective at enhancing the predictive accuracy of the models. The results from each competing modeling approach are compared and managerial implications to improve the SDR performance indicator in SPAS are provided.

Data Mining Framework to Optimize the Bid Selection Policy for Competitively Bid Highway Construction Projects

In competitive bidding in the United States, the lowest bid is frequently selected to perform the project. However, the lowest bidder may incur significant cost increases through change orders. For project owners to accurately estimate the actual project cost and to predict the bid that is close to the actual project cost, there is a need for new decision aids to analyze the bid patterns. In this paper, two neural network models, a classification model and a general regression model, were used as a method of selecting the bidder that submits the bid closest to the actual project cost. The empirical results suggest that for selected projects these models selected the bids that are closer to the actual project costs than the lowest bid. The outcome of this study addresses the issue of cost overrun, which is a very common problem in the construction industry.

ANALYZING BIDDING STATISTICS TO PREDICT COMPLETED PROJECT COST

Ratios were constructed relating the second lowest bid, mean bid, median bid, maximum bid to the low bid for highway construction projects in Texas to study if there are useful patterns in project bids that are indicators of the project completion cost. It was found that the value of the ratios tend to be larger for projects where the completed cost deviates significantly from the original low bid. Larger ratio values were found for projects that were completed with a greater than 20% cost increase and for projects that were completed with a reduction in cost from the original bid amount of 10% or more. Regression and neural network models were developed to predict the completed cost of Texas highway projects using the bidding ratio data. The input data used were transformed using the natural logarithm. Various combinations of the calculated ratios, as well as the project low bid were used as input to the models. The models tested produced predictions of varying accuracy. The models produced predictions that varied between having 41,02% and 54.01% of their test cases within 5% of the actual completed project cost. The model with the highest percentage of closely predicted cases was a regression model using the mean bid ratio, the coefficient of variation, and the low bid as inputs.

USING BIDDING STATISTICS TO PREDICT COMPLETED CONSTRUCTION COST

ABSTRACT The completed cost of a competitively bid construction project often exceeds the original low bid. This paper presents two models to predict completed construction cost based upon characteristics of the submitted bids. Data on completed projects were obtained from New Jersey Department of Transportation for 298 highway construction projects. Median bid and normalized median absolute deviation (NMAD) were selected from various bid characteristics as the best predictors of completed construction cost. Regression and neural network models were developed from the data. Both models have similar utility to predict completed costs. Due to ease of use, the regression model is preferred over the neural network model.