Conclusions and Recommendations

Abstract

This book has five objectives. The first one is to introduce a promising MARS procedure for numerical mapping. The second is to show the main advantages of MARS over other methods including the variety of neural networks, the extreme learning machine ELM, the linear genetic programming LGP, multi-expression programming MEP, standard genetic programming GP, and adaptive neuro-fuzzy inference system ANFIS, for complex data mapping in high-dimensional data. The third is to present some applications of MARS algorithm in big data geotechnical problems, such as the HP pile drivability analysis. The fourth is to demonstrate the procedures of MARS use, including the model development, model interpretation, and parametric sensitivity analysis. The last is to illustrate the modified MARS procedure for pattern recognition/classification (MARS_LR), such as the liquefaction assessment and the stability evaluation of the entry-type excavation.