Professional Documents
Culture Documents
Published Paper
Published Paper
net/publication/245136059
CITATIONS READS
62 467
3 authors, including:
All content following this page was uploaded by Min-Yuan Cheng on 19 July 2023.
Abstract
In the construction industry, sub-contractor's performance is a crucial factor in their awards of a new job by a general contractor. The objective
of this study is to improve the current practices for evaluating sub-contractors performance.
Drawbacks of current evaluation process are discussed firstly. The appropriateness for adopting the Evolutionary Fuzzy Neural Inference
Model (EFNIM) for improving the drawbacks is studied. A Sub-contractor Performance Evaluation Model (SPEM) is then developed by
employing the EFNIM. The effectiveness of the proposed SPEM is validated by performing case study of a real general contractor. Validation
results show that the proposed method accurately measures sub-contractor's performance enhancing the current practice of evaluation.
© 2006 Elsevier B.V. All rights reserved.
Keywords: Sub-contractor; Performance Evaluation; Artificial Intelligence (AI); Genetic Algorithms; Fuzzy Logic; Neural Networks
1. Introduction of two stages: primary score and final score. The primary score
is examined by field superintendents, whereas final score is
Construction is an industry completed by numerous sub- evaluated by managers of general contractor after primary
contractors. When a construction firm wins a bid, it usually has evaluation. The difference between primary and final scores is
to subdivide the won bid and sub-contract some portions of it evaluating perspective. Due to intrinsic ambiguity and difficult
out [1]. In practice, the general contractor cannot provide the formalization, evaluations are usually accomplished by evalua-
construction site with unlimited resources. To reduce costs, tor's experience and intuition [9]. An objective evaluation
most jobs required to complete a project are executed by sub- method is thus urgent for decision makers to make proper
contractors [2]. Due to the peculiarities of projects, dissimilar decisions.
sub-contractors need various labors, materials, machineries/ Kumaraswamy and Matthews [10] proposed a pro forma to
equipment, techniques, etc. to do a job. There may also be a assess sub-contractors [10]. The pro forma was constructed by
tardiness penalty cost if the contracted project cannot be eight items such as partnering experience, response to
completed before its due date. As a result, sub-contractors construction thoughts, quality awareness, etc. Sub-contractors
performances associate with success of construction projects were subjectively evaluated and averages were taken for
[3]. A well-performed sub-contractor can achieve jobs within multiple evaluators. Ekstrom et al. [11] developed a tool to
planed duration with anticipated budget and quality. In the evaluate bidders in architecture/engineering/construction (AEC)
contrary, a poor-performed sub-contractor results a defective by calculating a weighted rating based on source credibility
work and therefore consumes additional costs and completion theory [11]. Performance evaluation of sub-contractors that
time [4–6]. fluctuates according to the changing environment is inherently
To determine sub-contracting a job to which sub-contractor, complex, uncertain, and lack of complete data [12]. Artificial
general contractors use sub-contractors performance as a Intelligence (AI) is concerned with building computer systems
reference index [7,8]. In general, the evaluation process consists that solve problems intelligently by emulating the human
behavior, which provides a promising direction to solve the
problem [13].
⁎ Corresponding author. Tel.: +886 4 851 1888x2241; fax: +886 4 851 1270. Albino and Garavelli [9] applied Neural Networks (NNs) in
E-mail address: fpecount@yahoo.com.tw (C.-H. Ko). rating sub-contractors [14]. NNs originated from modeling the
0926-5805/$ - see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.autcon.2006.09.005
526 C.-H. Ko et al. / Automation in Construction 16 (2007) 525–530
Table 2
Data collection of sub-contractor performances
Sub Contract Factor number Final
contractor no. score
1 2 3 4 5 6 7 8 9 10 11 12
A A_01 6 6 4 6 6 4 6 6 6 4 8 8 72
A_02 6 6 6 8 6 6 4 6 6 6 6 8 76
A_03 6 6 6 6 6 6 6 6 6 6 8 8 76
A_04 8 6 6 6 6 6 6 8 6 6 6 8 80
A_05 8 8 6 6 6 6 6 8 6 6 8 8 86
B B_01 8 6 6 8 6 6 4 6 6 6 8 8 80
B_02 6 6 6 8 6 6 6 6 6 6 8 8 76
B_03 8 6 4 6 6 6 4 8 6 6 8 8 76
B_04 8 6 4 8 6 4 6 8 4 4 8 8 74
B_05 6 6 6 6 4 4 4 8 6 6 8 8 70
C C_01 6 6 6 4 4 4 4 4 4 4 6 6 56
C_02 6 6 6 6 4 4 4 6 6 4 6 6 62
C_03 6 6 6 6 6 4 6 6 4 4 6 6 66
C_04 6 6 6 6 6 4 4 6 6 6 6 8 68
Fig. 3. Procedure of Sub-contractor Performance Evaluation Model. D D_01 6 6 6 6 4 4 4 4 4 4 6 8 60
D_02 6 6 6 6 4 4 4 6 4 4 6 8 66
D_03 8 6 6 6 4 4 4 6 6 6 6 8 68
Drawbacks of current evaluation process are discussed as
D_04 6 8 6 4 6 4 6 6 4 4 6 8 66
follows: E E_01 8 6 6 6 6 4 6 6 4 4 8 8 70
E_02 6 6 6 8 6 4 6 6 4 4 8 8 72
• Repeated evaluation process E_03 6 6 6 8 6 4 4 6 6 4 6 6 68
E_04 6 6 4 6 4 4 4 6 4 6 6 8 66
F F_01 6 6 6 6 4 4 4 4 4 6 6 8 62
Although the evaluation consists of two stages, decisions are
F_02 6 6 6 6 4 4 6 4 4 4 6 8 66
made based on the final score. If the relationships between F_03 6 6 6 6 4 4 4 4 4 4 6 6 60
evaluation factors and final scores can be identified, the primary F_04 6 4 4 4 6 4 4 4 4 6 6 6 58
score can be eliminated. Thus, evaluation efforts can be saved. F_05 6 4 4 4 4 4 4 4 4 4 6 6 56
G G_01 8 6 6 6 6 6 6 6 6 6 6 6 76
G_02 6 6 4 8 6 4 6 6 6 6 8 8 74
• Independent factors
G_03 8 6 6 6 6 6 6 6 6 6 6 8 76
G_04 8 8 8 6 6 6 6 6 6 6 8 8 80
The current evaluation process assumes all factors are G_05 8 8 6 6 6 6 6 6 6 8 8 8 86
independent. As a result, the inter-dependency between factors G_06 8 6 6 6 8 6 8 8 8 6 8 8 88
is ignored. H H_01 8 8 8 8 6 6 6 8 6 8 8 8 86
H_02 8 8 6 6 6 6 6 6 6 6 8 8 80
H_03 8 8 6 6 6 6 6 6 6 8 8 8 80
• Even weights H_04 8 6 6 6 6 6 6 6 6 6 8 8 76
I I_01 6 6 6 6 4 4 6 6 4 4 6 8 66
The current evaluation process assumes an equal influence I_02 8 6 6 6 6 4 4 6 4 6 6 6 68
on all factors, which cannot represent the significance for a I_03 6 6 6 4 6 4 6 6 4 6 6 8 66
I_04 8 6 6 6 4 4 4 6 4 6 8 8 70
higher impacted factor.
I_05 6 6 6 6 4 4 6 6 6 6 8 8 70
I_06 8 6 6 6 6 6 6 6 6 6 8 8 76
• Subjective judgment J J_01 8 6 4 6 4 4 4 6 4 4 6 8 66
J_02 6 6 6 6 4 6 4 6 6 4 6 8 70
J_03 8 6 6 8 6 6 6 6 6 4 6 6 76
J_04 8 6 6 8 6 4 6 8 6 4 6 8 74
Table 1 J_05 8 8 6 8 6 6 6 6 6 4 6 8 76
Evaluation factors of labor sub-contract type K K_01 6 6 4 6 6 4 6 8 6 4 8 8 74
Factor number Evaluation factors K_02 6 8 6 6 6 6 6 6 6 6 6 8 76
K_03 6 8 6 6 6 6 4 6 6 6 6 8 76
1 Construction technique K_04 8 8 6 6 6 6 6 6 6 6 8 8 80
2 Duration control ability
L L_01 6 6 4 6 6 4 4 4 6 6 6 6 62
3 Corporative manner L_02 6 6 4 4 6 4 4 4 4 4 6 6 58
4 Material wastage L_03 6 6 4 6 4 4 4 4 4 6 6 6 62
5 Services after work completion L_04 6 6 6 6 6 4 6 6 4 6 6 6 66
6 Collaboration with other sub-contractors
L_05 6 6 6 6 4 4 6 6 6 6 6 6 68
7 Safety and protection M M_01 8 8 6 6 6 6 6 6 6 6 8 6 76
8 Tool usage habit (tools borrowed from contractor) M_02 6 6 4 8 6 4 6 8 6 6 6 8 74
9 Working space clearance M_03 6 8 6 8 4 6 6 6 6 4 8 8 76
10 Management ability
M_04 8 6 6 8 6 6 4 6 6 6 8 8 76
11 Sub-contractor's personality
12 Economic condition (continued on next page)
528 C.-H. Ko et al. / Automation in Construction 16 (2007) 525–530
Table 2 (continued )
Sub Contract Factor number Final
contractor no. score
1 2 3 4 5 6 7 8 9 10 11 12
N N_01 6 8 4 4 4 4 4 4 6 6 6 6 62
N_02 6 6 4 6 4 6 6 4 4 4 8 6 56
N_03 6 6 4 6 4 4 6 4 6 4 6 6 60
N_04 6 6 6 6 6 4 6 6 6 4 8 6 66
N_05 6 6 6 6 6 4 4 6 6 6 6 6 68
N_06 6 8 4 4 4 4 6 6 4 4 6 8 66
N_07 6 6 4 6 4 4 4 6 4 6 8 8 66
N_08 6 6 6 6 4 6 6 6 6 6 6 6 70
N_09 8 6 6 6 6 6 6 6 6 6 6 6 76
N_10 6 6 4 8 6 4 6 8 6 4 8 8 74
N_11 6 8 6 8 6 6 6 6 6 6 6 8 76
N_12 6 6 6 6 6 6 6 6 6 6 6 8 76
N_13 8 8 6 6 6 6 6 8 6 6 8 8 80 Fig. 4. Evolutionary process.
N_14 8 8 6 6 8 6 6 8 6 6 8 8 86
N_15 8 8 6 8 8 6 8 8 8 6 8 8 88
evaluation items and final score; and GAs for evolving the
Each evaluation factor is measured by subjective assessment. optimum solution for the problem as opposed to using a trial-
Besides, assessment discrepancy in evaluation is caused by and-error manner to develop a solution. In addition, the model
diverse backgrounds and working experiences. can drive off the subjective judgment in the evaluation process.
AI techniques have characteristics of learning and adapta- Therefore, the EFNIM is an appropriate tool to overcome the
tion, which is appropriate for improving current practice [21]. drawbacks of current evaluation process.
To overcome the drawbacks, this research adopts an Evolu- This research adopts EFNIM as a core engine to develop the
tionary Fuzzy Neural Inference Model (EFNIM), which fuses SPEM whose procedure is shown in Fig. 3. The SPEM is
GAs, FL, and NNs, to develop a Sub-contractor Performance developed base on current evaluation process. In the figure, two
Evaluation Model (SPEM). activities are removed from the current practice. The redundant
activity of “summarize evaluated scores” is removed since
3. Evolutionary Fuzzy Neural Inference Model (EFNIM) decisions are made based on final scores, whereas, “file
evaluation results” is removed as the EFNIM records inference
The EFNIM was developed by Ko in [20] based on FL that results in a database. To identify knowledge required to evaluate
mimics the high level of human inference process. The sub-contractor performance, EFNIM is used to evolve solutions
architecture, a synergism of GAs, FL, and NNs, is shown in from previous cases. The performance is finally evaluated using
Fig. 2. FL and NNs are complementary technologies. The evolved solution to reduce subjective measurement caused by
combination of these two technologies into an integrated system human intervention.
appears a promising path towards the development of intelligent
systems [22,23]. In the figure, fuzzy inference engine and fuzzy 5. Validation
rule base in traditional Fuzzy Logic Systems (FLS) are replaced
by the NN. Although the integration of FL and NN is more 5.1. Case study
reasonable than traditional FL to simulate the process of human
inference, the NN has demonstrated the difficulty in selecting an Effectiveness of the established SPEM is validated by a real
appropriate topology as well as appropriate parameters for a general contractor in Taiwan. The contractor was established in
network [24–26]. GA is frequently used for optimization [27].
Therefore, the EFNIM employs GA to simultaneously search
for the fittest shapes of Membership Functions (MFs), optimum
NN topology, and optimum parameters of NN. Details about
EFNIM can be found in [20].
6. Conclusions
[8] R.R. Ramirez, L.F.C. Alarcon, P. Knights, Benchmarking system for [18] K.C. Lam, Tiesong Hu, S. Thomas Ng, Martin Skitmore, S.O. Cheung, A
evaluating management practices in the construction industry, Journal of Fuzzy Neural Network approach for contractor prequalification, Con-
Management in Engineering, ASCE 20 (3) (2004) 110–117. struction Management and Economics 19 (2001) 175–188.
[9] V. Albino, A.C. Garavelli, A Neural Network application to subcontractor [19] L.C. Jain, N.M. Martin, Fusion of Neural Networks, Fuzzy Sets, and Genetic
rating in construction firms, International Journal of Project Management Algorithms: Industrial Applications, CRC Press, Boca Raton, Florida, 1999.
16 (1) (1998) 9–14. [20] C.H. Ko, Evolutionary Fuzzy Neural Inference Model (EFNIM) for
[10] M.M. Kumaraswamy, J.D. Matthews, Improved subcontractor selection decision-making in construction management, PhD thesis, National
employing partnering principles, Journal of Management in Engineering, Taiwan University of Science and Technology, Taipei, Taiwan, 2002.
ASCE 16 (3) (2000) 47–58. [21] S. Haykin, Neural Networks: A Comprehensive Foundation, Prentice-Hall,
[11] M.A. Ekstrom, H.C. Bjornsson, C.I. Nass, Accounting for rater credibility Upper Saddle River, New Jersey, 1999.
when evaluating AEC subcontractors, Construction Management and [22] A.M.P. Canuto, W.G.J. Howells, M.C. Fairhurst, Fuzzy multi-layer
Economics 21 (2) (2003) 197–208. perceptron for binary pattern recognition, Proceedings of the 1999 7th
[12] M.A. Ekstrom, H.C. Bjornsson, C.I. Nass, Accounting for rater credibility International Conference on Image Processing and its Applications, vol. 1,
when evaluating AEC subcontractors, Journal of Construction Manage- IEE, Stevenage, England, 1999, pp. 260–264.
ment and Economics 21 (2) (2003) 197–208. [23] S. Rajasekaran, G.A. Vijayalakshmi Pai, Simplified fuzzy ARTMAP as
[13] T.K. Wu, Performance evaluation and prediction model for construction pattern recognizer, Journal of Computing in Civil Engineering, ASCE 14
subcontractor, MS thesis, National Taiwan University of Science and (2) (2000) 92–99.
Technology, Taipei, Taiwan, 2001 (in Chinese). [24] D.P. Kanungo, M.K. Arora, S. Sarkar, R.P. Gupta, A comparative study of
[14] V. Albino, A.C. Garavelli, A neural network application to subcontractor conventional, ANN black box, fuzzy and combined neural and fuzzy
rating in construction firms, International Journal of Project Management weighting procedures for landslide susceptibility zonation in Darjeeling
16 (1) (1998) 9–14. Himalayas, Engineering Geology 85 (2006) 347–366.
[15] A.I. Wu, P.K.S. Tam, A simplified model of fuzzy inference system [25] G. Leng, M. Thomas Martin, G. Prasad, An approach for on-line extraction
constructed by using RBF neurons, Proceedings of the 1999 International of fuzzy rules using a self-organising fuzzy neural network, Fuzzy Sets and
Conference on Fuzzy Systems, vol. 1, 1999, pp. I-50–I-54. Systems 150 (2) (2005) 211–243.
[16] M.I. Okoroh, V.B. Torrance, A model for subcontractor selection in [26] H. Iyatomi, M. Hagiwara, Adaptive fuzzy inference neural network,
refurbishment projects, Construction Management and Economics 17 (3) Pattern Recognition 37 (10) (2004) 2049–2057.
(1999) 315–328. [27] M. Gen, R. Cheng, Genetic Algorithms and Engineering Design, Wiley,
[17] L.A. Zadeh, Fuzzy sets, Information and Control 8 (3) (1965) 338–353. New York, 1997.