Pre-processing and Input Vector Selection Techniques in Computational Soft Computing Models of Water Engineering

Input feature selection has a crucial role in predictive computational soft computing models. This chapter explored the appropriate pre-processing techniques and input vector selection methods for soft computing models. The pre-processing techniques, namely principal component analysis (PCA), Boruta feature selection algorithm (BFS), the gamma test (GT) algorithm, and subset selection by maximum dissimilarity (SSMD) algorithm, in the field of soft computing models is introduced, and implemented in bedload transport predictions, as a test case. The results of the current study highlighted the effectiveness of pre-processing, input variable selections, determination of the dominant input features and provide significant practical reference value for soft computing model developments.