A multiple kernel learning-based decision support model for
contractor pre-qualification
K.C. Lam
a
, C.Y. Yu
b,
⁎
a
Dept. of Building and Construction, City Univ. of Hong Kong, Hong Kong SAR
b
Dept. of Building and Construction, City Univ. of Hong Kong, Hong Kong SAR
abstractarticle info
Article history:
Accepted 4 November 2010
Available online 3 December 2010
Keywords:
Contractor pre-qualification
Decision support model
Support vector machine
Multiple kernel learning
Due to the complex nature of the contractor pre-qualification such as subjectivity, non-linearity and multi-
criteria, advanced model should be required for achieving a high accuracy of this decision-making process.
Previous studies have been conducted to build up quantitative decision models for contractor pre-
qualification, among them artificial neural network (ANN) and support vector machine (SVM) have been
proved to be desirable in solving the pre-qualification problem with regards to their higher accuracy and
efficiency for solving the non-linear problem of classification. Based on the algorithm of SVM, multiple kernel
learning (MKL) method was developed and it has been proved to perform better than SVM in other areas.
Hence, MKL is proposed in this research, the capability of MKL was compared with SVM through a case study.
From the result, it has been proved that both SVM and MKL perform well in classification, and MKL is more
preferable than SVM, with a proper parameter setting. Therefore, MKL can enhance the decision making of
contractor pre-qualification.
© 2010 Elsevier B.V. All rights reserved.
1. Introduction
Contractor pre-qualification is a process in which a pool of con-
tractors is first screened based on a set of criteria, after which the
shortlisted contractors shall be invited to join the subsequent bid-
ding process. Pre-qualification is critical to assure that the selected
contractors would all be capable to carry out their contracted work (as
prescribed in the specifications, drawings, etc.) satisfactorily if the
contract is awarded to them. This screening process may help public
or private sectors to efficiently apply their funds by ensuring that a
qualified contractor has been selected to execute the work [1]. With
the implementation of the contractor pre-qualification exercise, the
subsequent bidding process can be focused on price only and thus
allows a fair competition among the bidders, which shall lead to the
best price for the client.
The decision of contractor pre-qualification was regarded as a com-
plicated two-group non-linear classification problem [1]. It involves
some problems such as subjectivity, non-linearity, and multi-criteria.
Subjectivity means that the decisions are depending on the intuitive
judgments of the decision makers; non-linearity may be caused by
the non-linear relationship between the score of the individual crite-
rion and its impact to the decision to be made, as well as the different
weights borne by the individual criterion; for multi-criteria, it reflects
the existence of a large number of decision criteria that may or may not
be at the same level [1–3]. In practice, the contractors' suitability to
participate in a project bid is usually assessed by the project owners in
accordance with their previous experience, judgment and a set of cri-
teria which might vary between projects and clients. Due to the com-
plex nature of the contractor pre-qualification decision-making process,
comprehensive models should be required.
In the past decade, extensive research effort has been made to cope
with the non-linearity existing between contractor attributes and the
corresponding pre-qualification decisions made by the owner–client.
These effort include the development of non-linear or reference mod-
els; knowledge-based expert systems or Qualifer-2 [4]; case-based
reasoning [5]; and the artificial neural network (ANN) models [2,3],
they were also employed by many researchers to aid the clients in
dealing with the inherited non-linearity, uncertainty and imprecision
involved in the process of contractor pre-qualification and selection.
During the last few years, kernel methods, such as SVM have been
proved to be powerful for a wide range of different data analysis
problems, and be efficient tools for solving learning problems like
classification [6]. Lam et al. [7] proposed a support vector machine
(SVM)-based model for contractor pre-qualification, it is proved that
SVM can obtain relatively desirable results compared with ANN.
However, recent developments in the literature [7] on SVM and other
kernel methods have shown the need to consider multiple kernels.
Hence, the multiple kernel learning (MKL) method is proposed in this
r
esearch, to see if it can perform more desirably than SVM towards the
contractor pre-qualification problem.
Automation in Construction 20 (2011) 531–536
⁎ Corresponding author: Tel.: +852 2194 2743; fax: +852 2788 7612.
E-mail addresses: bckclam@cityu.edu.hk (K.C. Lam),
chenyunyu3@student.cityu.edu.hk, sharonyun1202@yahoo.com.cn (C.Y. Yu).
0926-5805/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.autcon.2010.11.019
Contents lists available at ScienceDirect
Automation in Construction
journal homepage: www.elsevier.com/locate/autcon