Metaheurestic algorithm based hybrid model for identification of building sale prices

Abstract

The overall cost of a building depends on several variables such as economical, project physical and financial variables. The CCB (construction cost of building) also depends on deviations of several indices which are not control in an easy way. Therefore, the overall sales prices of a building may not be controlled due to these indices. In this chapter, a metaheuristic algorithm based hybrid model for identification of building's sales prices is presented, which is developed by using conventional Feedforward Neural Network (FNN).table The identification accuracy of FNN is varies with respect to the number of input variables and its modal parameters such as weight (w) and bias (b). In this chapter, the number of most relevant input variables are selected by using Relief F Attribute evaluator (RFAE) with the help of ranker search method. After selecting most appropriate variables, the FNN parameters are optimized by using particle swarm optimization (PSO) based metaheuristic algorithm (MA). The total 208 intelligent models have been designed and validated using 372 real side construction cost dataset of three to nine story buildings. The validated results by FNN and PSO-FNN show that selected variables gives better results as compared with other models.