Predicted Cost Escalations in Competitively Bid Highway Projects.

Predicting project cost overrun levels in bidding stage using ensemble learning

“…They seek more insight into the meaning of input data since the ensemble learning method is in the earlier stage of the construction cost prediction area. Williams, Miles, and Moore (1999) indicated that a strong linear relationship exists between the natural logarithm of a low bid and that of the completed cost for highway projects in Great Britain and the United States. Skitmore and Ng (2003) developed different forms of regression models to estimate the actual construction cost and time.…”

“…Project type (sample size) Input data Jahren and Ashe (1990) Naval facilities (US, 1576) Project size, the difference between the low bid and government estimate, the type of construction, and the level of competition Williams, Miles, and Moore (1999) Highway (UK,28,US,90) Low bid Williams (2002Williams ( , 2005 Highway (NJ, 302); Highway (TX, 1260) Low bid, median bid, expected project duration, and number of bids Attalla and Hegazy (2003) Reconstruction project (Canada, 50) 36 variables (scope definition and planning, tendering stage, schedule, cost, quality, communication, and safety) Skitmore and Ng (2003) Australian construction projects (various, 93) Client sector, contractor selection method, contractual arrangement, and project type Ling, Chan, and Chong (2004) Residential (Singapore, 87) 59 variables, delivery methods (DB/DBB) Georgy, Chang, and Zhang (2005) Industrial construction projects (US,50) 25 variables (project size, contract type, relative level of complexity, site conditions, and design schedule) Gkritza and Labi (2008) Highway (Indiana, 1957) Project type, project duration, and contract size Son, Kim, and Kim (2012) Commercial buildings (84) 64 variables (pre-project planning stage: project type, project size, and project duration) Williams and Gong (2014) Highway (California, 1221) Low bid, the completed project cost, and the number of bidders. and bidding text documents Sousa et al (2014) Sanitation (Chicago, 180) Delivery method (DBB), and project type (water/sewer) Table 4.…”

“…The difference between the two studies stems from available empirical data. Most previous study related to bidding stage prediction use only bidding characteristics (e.g., bid price, the number of bidders) (Williams, Miles, and Moore 1999;Williams 2002Williams , 2005Williams , 2007Williams and Gong 2014). However, this study used bidding characteristics combined with project characteristics, thereby resulting in an improvement in prediction accuracy.…”

The Regulatory Asset Base Model and the Project Finance Model: A comparative analysis

“…On the other hand, it enables strategic behaviour, where the bidders conceal their true revenue expectations during the bidding stage (Liu et al 2007), leading to pressures on the sponsor during the construction. Indeed there is a documented relationship (Williams et al 1999;Jahren and Ashe 1990) between the magnitude of the low bid and measured cost overruns (against the winning bid).…”

Public Private Partnerships for Transport Infrastructure :

“…(Flyvbjerg et al 2002) and manifests itself in scope creep during project execution. Opportunistic behaviour on the contractor side manifests a well-known functional relationship between the distance of low (winning) bids from the average of other bids and the magnitude of later cost overruns, recovered through renegotiation (Williams et al 1999;Jahren and Ashe 1990). …”