Analysis of the perception of relative importance of critical success factors in the civil construction industry using artificial neural networks

In engineering an efficient project management can guarantee success and reduce the impacts of delays and changes that occur during the execution of the enterprise. It is necessary to distinguish company practices essential to success and the alignment of resources. To do so, there are the Critical Success Factors (CSF), which identify the set of key areas that are essential for the company to achieve its mission. With the identification of CSF, the more efficient the measures to be taken, avoiding unsuccessful projects. The objective was to create a model for determining the CSF that affect Project Management in the construction industry with the use of Artificial Neural Networks. The methodology was based on the analysis of documents using PRISMA method (checklist with 27 items) where only documents that were considered relevant to the study and interviews remained. After identifying the potential Success Factors (SF) that affect Project Management, a questionnaire was prepared to evaluate the effect of each factor. The analysis of the answers was made by the Relative Importance Index (RII). ANN was used to evaluate the most significant success factors using the Neuro4 software with the Resilient Propagation algorithm for optimization of processes to obtain satisfactory ANN. A total of 37,822 divided articles and 2,328 journals were searched, with the following terms: Critical Factors and Project Management, Critical Factors and ANNs, Project Management and ANNs and Project Management and Critical Factors and ANNs. Of 874 e-mails sent to the Institutions of Education registered in the e-mec system, 191 answered distributed in the 27 Brazilian states, where 70% occupy the position of professor/researcher, 26% coordinators, 2% Rector and 1% of Direction/Management. The relative indexes of importance were calculated and the matrix that was used in the Neuro4 was obtained to obtain weights for the determination of the CSF using the Garson Algorithm. The factor: Unrealistic inspection and test methods proposed in the contract was the most critical in the management of projects in the two times considered, Research Time and Experience Time of the respondents. The ANNs produce subsidies to know the relevance of the input variables adopted, are efficient to order and transfer unstructured knowledge and constitute a fast and precise means for modeling nonlinear variables.