Predicting bid prices in construction projects using non-parametric statistical models

Bidding is a very competitive process in the construction industry; eachcompetitor’s business is based on winning or losing these bids. Contractors would like topredict the bids that may be submitted by their competitors. This will help contractors toobtain contracts and increase their business. Unit prices that are estimated for eachquantity differ from contractor to contractor. These unit costs are dependent on factorssuch as historical data used for estimating unit costs, vendor quotes, market surveys,amount of material estimated, number of projects the contractor is working on,equipment rental costs, the amount of equipment owned by the contractor, and the riskaverseness of the estimator. These factors are nearly similar when estimators areestimating cost of similar projects. Thus, there is a relationship between the projects thata particular contractor has bid in previous years and the cost the contractor is likely toquote for future projects. This relationship could be used to predict bids that thecontractor might quote for future projects. For example, a contractor may use historicaldata for a certain year for bidding on certain type of projects, the unit prices may beadjusted for size, time and location, but the basis for bidding on projects of similar typesis the same. Statistical tools can be used to model the underlying relationship between the final cost of the project quoted by a contractor to the quantities of materials oramount of tasks performed in a project. There are a number of statistical modelingtechniques, but a model used for predicting costs should be flexible enough that it couldadjust to depict any underlying pattern.Data such as amount of work to be performed for a certain line item, materialcost index, labor cost index and a unique identifier for each participating contractor isused to predict bids that a contractor might quote for a certain project. To perform theanalysis, artificial neural networks and multivariate adaptive regression splines are used.The results obtained from both the techniques are compared, and it is found thatmultivariate adaptive regression splines are able to predict the cost better than artificialneural networks.