Implementation of data mining techniques in construction estimating

  • Authors:
  • Zeljko Popovic

  • Affiliations:
  • Parsons Brinckerhoff (PB Power Ltd), Middle East Regional Office, PO Box 47445, Abu Dhabi, United Arab Emirates

  • Venue:
  • Neural, Parallel & Scientific Computations - Special issue: Computing intelligence in management
  • Year:
  • 2004

Quantified Score

Hi-index 0.00

Visualization

Abstract

Constructions cost estimating databases are utilized for definition of contract price and evaluation of construction progress and claims. They are known for standardized data format and procedures, involving pre-defined coding system, pointing to ready-made database items and printing of standard reports, such as: bills of quantities.Abundance in data does not always result in optimal estimate and improvements to the estimating technology are often sought. Improvements should not dramatically change the basic cost model, widely accepted in practice, arranged in a convenient manner per type of work, and incorporated into various contract documents.The most effective improvements were identified and proposed in two major areas: estimating procedure and cost modeling. Estimating procedure was enhanced to include different data mining techniques, including interactive visualization, research/comparison of historical costs, and simultaneous usage of several cost models derived from basic data. Cost data model was further extended to incorporate fuzzy logic and "flexible queries", as to simulate experts reasoning and provide automated and "intelligent" professional advice to the user.Proposed improvements were implemented using combination of standard relational database language and object-oriented programming and then tested on a large construction project. Results matched experts' opinion and "intelligent database" proved to be useful tool in estimating practice.