SVM和RLR算法的对比分析

Comparative Analysis of SVM and RLR Algorithms

摘要: 在机器学习领域中，支持向量机(SVM)和逻辑回归(RLR)作为两种有监督的分类算法，在不同场合下有着不同的分类效果。本文旨在通过数据分析对二者进行比较。数值实验结果显示：随着数据样本维度的增加，SVM的预测准确率、稳定性、计算时间及计算资源占用情况比RLR更好；对存在离群值的样本数据分类时，SVM在稳定性和分类效果方面表现更佳；在高维小样本中，RLR预测准确率比SVM更高，表现更佳。

Abstract: In the field of machine learning, support vector machine (SVM) and logistic regression (RLR), as two supervised classification algorithms, have different classification effects in different situations. This paper aims to compare them through data analysis. The numerical experiment results show that, with the increase of data sample dimension, SVM has better prediction accuracy, stability, calculation time and calculation resource occupation than RLR. When classifying the sample data with outliers, SVM performs better in terms of stability and classification effect. In high-dimensional small samples, RLR has higher prediction accuracy and better performance than SVM.