we need to have a prediction
According to Beatham et
al (2004) the present practice of project success/failure measurement
encourages the measurement of project performance with “lagging indicators” and
leads us to expect project “autopsy reports”. This, however, does not offer
opportunity for change and improvements as expected from assessment in early
time. the concept of organizational learning could be of benefit to the
on-going project, and if lessons learned from a completed project could provide
a guide for future projects, then it is the case that assessment should cover its
entire “life story”. The question here is, whether the success or failure of a
project is of any relevance to the project after they had occurred? To correct these, such measurements should
always be aimed at giving opportunities to change and, always leading to enhancements
the cost of a construction project is a task that is critical in the project’s
success. Therefore, a cost prediction model based on limited information in the
early phase of a construction project is essential. The same importance should
give also, to the time prediction. Many studies produced for the UK Government
have highlighted inaccurate prediction of client costs and construction
duration period as key problems for the construction sector.
Researchers have made
various attempts to develop a prediction model for construction cost. An
analysis of these earlier studies showed the following three important issues
(Arditi, & Günaydin, 2008; MOCT, 2008).
While some of the existing
prediction models are applied to the design phase, most of them can only be
applied to the feasibility phase, the initial stage of a project.
Prediction models predict mainly
the total construction cost, but the data on which the result is based cannot
The relationships between the
attributes used to predict the construction cost and the characteristics of the
construction project, which affect the construction cost, have yet to be
For strategists, management
need to predict the KPIs future values to make good decisions during the design
and the determination of the suitable target for each objective and KPIs. From
historical data, the dependency between KPIs variables can be discovered
through developing prediction model. Hence, the KPI future values can be
Technique of Construction Prediction Models
Regarding the selection
of predictive technique, the field of predictive analytics offers a verity of
techniques from statistics, data modeling, and machine learning that can analyze
historical facts to make predictions about the future performance of projects.
However, due to the amount of available data and conceptual stage of the
research, the techniques that can explain their reasoning of prediction are
preferred. The previous researches used several methodologies to solve the
problem of predicting the construction projects variables and KPIs. Some of the
methods used in the previous studies include:
Statistical methods such as
multiple regression analysis (MRA).
Based on the preliminary screening, the linear multi regression models seemed
to be reasonable choice. They are simple and often provide an adequate and
interpretable description of relationship between inputs and outputs. (Hastie
et al. 2008)
Repetitive learning methods such as
networks (ANN) for
predicting construction variables.
Stochastic methods such as Monte-
Carlo simulation (MCS).
Analogical methods such as
Case-based reasoning (CBR). CBR is a data-mining technique that recalls similar
situations applied to the solution of previous problems, and uses the
information and knowledge from such situations to solve a new problem. This
technique does not require a clear model for problem-solving; rather,
establishing cases is an important task in problem-solving.
The next section gives further
details about the common used methods from the above four methods to create