Estimating water treatment plants costs using factor analysis and artificial neural networks Mohamed Marzouk * , Mostafa Elkadi Structural Engineering Department, Faculty of Engineering, Cairo University, Egypt article info Article history: Received 16 January 2015 Received in revised form 3 September 2015 Accepted 5 September 2015 Available online 25 September 2015 Keywords: Parametric cost estimation Water treatment plant Cost driver Descriptive statistics ranking Principal component analysis Artificial neural networks abstract The cost of construction project is a fundamental input for decision making process set by owner during procurement stage. The paper identifies the cost drivers that are used in parametric cost estimation model for water treatment plants. Cost estimation at planning stage of projects is important for the success of the next stages in the projects. It is also very useful at the design stage of a project when information is incomplete and detailed designs are limited in such an early stage. Literature has been reviewed and interviews were conducted with experts and officials in water treatment plants to explore all variables that influence the construction cost of water treatment plants. A questionnaire survey was then conducted to assess the impact of the identified factors on construction costs of water treatment plants. Datasets that consist of 160 water treatment plant projects in Egypt were collected. Construction cost drivers have been nominated through two different procedures. The first technique is descriptive statistics ranking of variables by evaluating Mean Score and Relative Importance Index based on re- spondents' feedback in conducting questionnaire. The second technique utilizes exploratory factor analysis on the collected dataset. These cost drivers are used to construct two predictive models for estimating the construction cost of water treatment plants models using artificial neural networks. Analysis of results was performed to validate the models and demonstrate their effectiveness. The proposed methodology aids public authorities to perform comparative analysis and evaluate the different alternatives of water treatment plant projects. © 2015 Elsevier Ltd. All rights reserved. 1. Introduction Water treatment plants are classified as infrastructure projects that are usually administrated by public authorities. Such class of projects is critical and has different components including build- ings, underground piping and equipment. In recent decades, Egypt witnessed construction of many water treatment plants. Early stage cost estimate plays a significant role in any initial water treatment plants projects decisions, despite the project scope has not yet been finalized and still very limited information regarding the detailed design is available. Major problems faced are lack of preliminary information, lack of database of water treatment plants construction costs, missing data, lack of appropriate cost estimation methods, and the involvement of uncertainties. Water treatment plants' stakeholders in Egypt often need to estimate the construction costs of these plants at early stage readily and approximately to secure the required fund. Therefore, it is important to find a reasonable cost estimate tool for water treat- ment plants projects with acceptable level of accuracy. In Egypt, the estimating variability is high and can reach an average range of À37.8% to þ28.56 within a contractor's company (Khorshid and Abdel-Razek, 1991). Due to the observed high value of estimating variability, this research was carried to develop a fast and reliable model using artificial neural networks. Cost estimation is a heavily experience-based process that in- volves the evaluation of several complex relationships of cost- influencing factors, largely based on professional judgment (Alex et al., 2010). Parametric cost estimations technique is imple- mented in the early stage of a project. According to Project Man- agement Body of Knowledge (PMBOK, 2013), parametric estimating is defined as a technique using a statistical relationship between Abbreviations: EFA, Exploratory Factor Analysis; ANN, Artificial Neural Network; EFCBC, Egyptian Federation for Construction and Buildings Contracts; PCA, Principal Component Analysis; MSA, Measures of Sampling Adequacy; KMO, Kaiser-Meyer- Olkin; MLP, Multilayer Perceptron; SPSS, Statistical Package for the Social Sciences. * Corresponding author. E-mail address: mm_marzouk@yahoo.com (M. Marzouk). Contents lists available at ScienceDirect Journal of Cleaner Production journal homepage: www.elsevier.com/locate/jclepro http://dx.doi.org/10.1016/j.jclepro.2015.09.015 0959-6526/© 2015 Elsevier Ltd. All rights reserved. Journal of Cleaner Production 112 (2016) 4540e4549