Using Artificial Neural Networks to Model Cost Overruns in Real Estate Projects
- November 2, 2018
- Posted by: RSIS
- Category: Civil Engineering
International Journal of Research and Scientific Innovation (IJRSI) | Volume V, Issue X, October 2018 | ISSN 2321–2705
Using Artificial Neural Networks to Model Cost Overruns in Real Estate Projects
Kareem Mohammad Mahmoud Mostafa, Dr. Ayman Hamdy Nassar
Civil Engineering Program, Faculty of Engineering and Material Sciences, The German University In Cairo, Cairo, Egypt
Abstract: – Cost overrun is one of the most important problems and risks that encounter Real Estate projects success, since it reduces the contractor’s profit and sometimes lead to enormous losses, and leaving the project in great troubles. Construction cost is one of the peak criteria of success for a real estate project throughout its lifecycle and is of high concern to those who are involved in the real estate industry. All real estate projects, regardless of their size, complexity are saddled by targets and uncertainties. Mostly in developing countries real estate projects are characterized by overruns in cost. Cost overruns occur in every real estate project while the magnitude varies significantly from project to project. This leads to severe need of addressing the acute issue of cost overrun.
Throughout this research we tried to gather and analyze the main factors which cause cost overruns in real estate projects. According to past researches the factors determining percentage overrun include the financial condition of the owner , the cash flow of the contractor, material cost increase, competition at tender stage, fluctuation in the currency that payments will be made by, the project size, delays in design approval, quantity variations, the detailed degree of the drawings used for estimating the budget, , material estimate accuracy, quality requirements being hard to reach, design changes, location of the project with respect to vendors, time needed for decisions to be made, what was known about the project at tender stage, the client’s characteristics, unknown geological conditions, ignorance and lack of knowledge of the parties, suitability of the project schedule to the project, conflict among participants in the project, design complexity, scope change by the owner, incompatible advanced payment, the prequalification of the contractor, workload in the project, the contract type , whether the parties agreed on dispute settlement procedure or not, the inspection and testing procedures , whether the site was properly managed or not , the adequacy of the equipment used in the project, the adequacy of the safety procedures followed, the experience of the contractor in similar projects, site access ease by the contractor, the effectiveness of the planning and scheduling , the availability of equipment, delay in arrival of material, shortage of labors, whether the labors used were skilled or not, the adequacy of the method of construction, the decreased productivity of labors and equipment, the availability of cost control engineers assigned for the project, the category of the contractor and finally the method of procurement.
Two Questionnaires were created to determine the probability and impact of each factor and hence rank the factors according to Relative importance index besides determining the percentage overrun each factor could cause according to practitioners in the field of real estate. Not to mention, data of real projects which encountered overruns were included. Using Artificial Neural networks which is a branch of Artificial Intelligence, a model was developed including the 43 factors which can predict the percentage overrun for real estate project. The model was later tested by one of the projects and the variance between the actual overrun and the predicted overrun was calculated.
