基于支持向量机算法的数据分类验证研究

Data Classification Based on SVM Algorithm Validation Studies

支持向量机是利用已知数据类别的样本为训练样本,寻找同类数据的空间聚集特征,从而对测试样本进行分类验证,通过验证可将分类错误的数据进行更正.以体检数据为数据背景,首先通过利用因子分析将高维数据进行降维,由此将所有指标整合成几个综合性指标;为降低指标之间的衡量标准所引起的误差,利用MATLAB软件将数据进行归一化处理,结合聚类分析将数据分类;最后利用最小二乘支持向量机分类算法进行分类验证,从而计算出数据分类的准确率,并验证了数据分类的准确性和合理性.

Support vector machine （SVM） is the use of data whose category label has been known as training samples, which is characterized by finding spatial aggregation of similar data to verify the classification of the test sample to correct the classification error data. Based on the health examination data for background in this paper, we choose factor analysis to reduces its dimensionality and all the indicators can be integrated into a few comprehensive indexes. To reduce the error caused by indicators, using the MATLAB to normalize the data and combine with method of clustering analysis to divide it into different health category. Finally, this paper, by using the least squares support vector machine （SVM） classification algorithm to classify validation, so as to calculate the accuracy of data classification and verify the accuracy of data classification and rationality.