Atmospheric Corrosion Modelling with SVM Based Feature Selection

Atmospheric corrosion has caused more and more losses and costs these years, so the world begin to pay much attention to this problem. In this paper, we mainly discuss the feature selection of a small subset of several important environmental factors from many relevant ones. With our experimental data with very small sample size, a model of corrosion rate is built. After specialized data preprocessing, we introduce the novel method of feature selection based on support vector machine (SVM). We demonstrate experimentally that the factors selected by this algorithm yield better modelling precision. Least square (LS) based feature selection method is also included in our experiment to reveal the superiority of SVM algorithm in our problem.