应用监督局部线性嵌入算法的科技项目质量评价

Evaluation of the Quality of Technology Projects Based on Supervised Locally Linear Embedded Algorithm

针对科技项目管理指标的数据维度高且相互影响而呈现出的复杂非线性关系为准确评价和科学管理带来挑战的现状,同时考虑到传统数据降维算法大多对非线性数据映射效果较差,采用监督局部线性嵌入算法,通过数据样本类别信息修改距离公式进行特征维数计算以获得科技项目的真实低维数据。实验结果表明:与传统算法相比,该算法预处理的样本在分类方面具有较高的准确率。

Dimension of indicators data about technology project is the higher, and has interaction influence, and shows complex nonlinear relationship, which brings challenges for the accurate evaluation of scientific management. Considering that most of traditional reduction algorithms about data dimension are poor for nonlinear effects of mapping data, so that we used supervised locally linear embedding algorithm to modify the distance formula according to data sample classification information, finally we calculated the feature and got the real low-dimensional data. The experimental results show that compared with the traditional algorithm, the sample data preprocessed algorithm has a higher accuracy on the performance of classification.