Genetic Algorithms And Projection Pursuit Classification Model To Bidding Evaluation Decision-making

Bidding for construction projects is an international method widely used in the construction of a transaction. Its prominent feature is introducing an open, fair and just market mechanism into the construction field. It enables the party issuing the contract of the construction project, which includes contracting parties, intermediaries and other market players, to be more standard and safe in the activities. It can prevent corruption in the field of construction, ensure the quality of the project effectively, and improve returns on investment of the construction significantly.At present, the system of construction project bidding, particularly in the system of the engineering construction bidding, is not ripe in our country. Project evaluation methods are not uniform and science in various places. There is a big element of subjectivity and human disturbance, because the project evaluation model involved more weight matrix is given by some experts depending on experience.For this reason, this paper combines and applies the projection pursuit algorithm theory and the theory of genetic research into decision-making in the bidding for construction projects, in order to provide a new approach for the project evaluation.Firstly, for the establishment of index system, this paper introduces some concepts and traditional evaluation methods in the construction project bidding, and systematically analysis the project tender indicators affecting decisions. Secondly, we rationally transform the qualitative indicators affecting the project tender in the decision-making into quantitative ones. Thirdly, we use the projection pursuit classification evaluation model (PPC) to transfer the high-dimensional data into lower dimensional sub-space. Fourthly, we use real-coded accelerating genetic algorithm (RAGA) to optimize the projection index function and the model parameters. Then the decisions can be made with the size of the value of pursuit in the project. At last, we verify the model with examples, which indicate that the model is simple, efficient, feasible, and is possible to avoid human disturbance in the fuzzy comprehensive evaluation and other evaluation methods.