基于改进KNN-DPC算法的科技创新人才分类研究

Research on Classification of Technological Innovation Talents Based on Improved KNN-DPC Algorithm

为了提高科技创新人才培养过程中人才层次分类的效果,提出了一种结合主成分的改进K近邻优化的密度峰值聚类算法(IKDPC)。首先,论文将主成分分析思想及流程融入到K近邻优化的密度峰值聚类算法(KNN-DPC)中来提高对高维数据的处理能力;进而,为了克服复杂数据集和噪声点对KNN-DPC算法的影响,对局部密度度量方法进行了改进,并设计了全新的两种样本数据点的分配策略,从而有效提高了聚类效率和聚类质量;最后,将IKDPC算法针对科技创新人才样本指标数据进行实例研究,实证结果表明该算法能有效地对科技创新人才进行分类,并为科学合理地探究科技创新人才培养过程中的分类问题提供科学量化参考。

In order to improve the effect of talent level classification in the training process of technological innovation talents(TIT),a new density peak clustering algorithm based on the improved K nearest neighbor optimization(IKDPC)is proposed.Firstly,the idea and process of principal component analysis are integrated into the density peak clustering algorithm(KNN-DPC)optimized by K nearest neighbor to improve the processing ability of high-dimensional data.Then,in order to overcome the influence of complex data sets and noise points on the KNN-DPC algorithm,the local density measurement method is improved.The new two sample data points allocation strategy is designed to effectively improve the clustering efficiency and clustering quality.Finally,the IKDPC algorithm is applied to the sample data of scientific and technological innovation talents to carry out an example study.The empirical results show that the algorithm can effectively classify the TIT and provide scientific quantitative reference for scientific and rational exploration of the classification problems in the training process of the TIT.