Abstract: A method based on Support Vector Machine optimization by Particle Swarm Optimization and Genetic Algorithm is proposed for predicting bankruptcy. The proposed method integrates the merits of Particle Swarm Optimization, Genetic Algorithm and Support Vector Machine, which simultaneously searches optimal regularization parameter and kernel parameter of Support Vector Machine for optimal prediction model. A sample dataset comprised of bankruptcy and non-bankruptcy data derived from the UCI machine learning repository is used. The data are randomly read from the dataset and automatically preprocessed by normalization. A 7-fold cross-validation test is used to objectively evaluate the prediction results. The simulation results indicate that the proposed method can automatically and efficiently construct optimal Support Vector Machine. Compared with other methods, the proposed method has better generalization capability, faster learning speed and better bankruptcy prediction accuracy than the other methods.

Key words: Particle Swarm Optimization（PSO） algorithm, Genetic Algorithm（GA）, Support Vector Machine（SVM）, optimization, parameter, bankruptcy prediction

摘要： 介绍了一种基于粒子群算法和遗传算法优化支持向量机预测破产的方法。这种方法融合了粒子群算法、遗传算法和支持向量机诸多优点，并行地搜寻支持向量机最优的正则化参数和核参数，由此构建优化的预测模型。采用源自UCI机器学习数据库的破产和非破产混合样本数据集，随机地读入数据和进行数据预处理，运用7重交叉校验方法客观地评价预测结果。仿真结果显示，这种方法能自动有效地构建优化的支持向量机，与其他方法比较，具有更强的推广能力和更快的学习速度，而且具有更好的破产预测准确率。

关键词: 粒子群算法, 遗传算法, 支持向量机, 优化, 参数, 破产预测