基于融合改进K-means聚类算法的数据检测技术

Data detection technology based on fusion improved K⁃means clustering algorithm

针对现有医疗财务数据分析系统平台老旧,采用传统K-means算法进行数据处理时性能较差的问题,文中设计了一种财务异常数据检测算法。对于传统K-means算法存在的分类效果不佳、运行效率偏低等不足,该算法结合密度峰值法对样本点的局部密度和高密度距离进行计算,进而优化簇中心的选择。同时融合PCA降维算法减少了数据的冗余信息,进一步提高了运行效率。通过引入LOF离群检测算法对分簇后的数据进行检测,从而得到异常数据结果。实验测试中,所提算法在人工数据集上的平均ARI指标为0.844,真实数据集的准确率则达到了79.2%,在所有对比算法中均为最优,表明该算法具有良好的性能,可以对财务异常数据进行准确地检测。

The platform of the existing medical financial data analysis system is old,and the traditional K-means algorithm is usually used for data processing with poor performance.To solve this problem,this paper designs a financial anomaly data detection algorithm.For the shortcomings of the traditional K-means algorithm,such as poor classification effect and low operating efficiency,this algorithm combines the density peak algorithm to calculate the local density and high-density distance of sample points,and then optimizes the selection of cluster centers.PCA dimension reduction algorithm is used to reduce data redundancy information and improve operating efficiency.The LOF outlier detection algorithm is introduced to detect the clustered data,and finally the abnormal data results are obtained.In the experimental test,the ARI index of the proposed algorithm in the artificial dataset has reached 0.844,and the accuracy rate of the real dataset has reached 79.2%,which is the best among all comparison algorithms,indicating that the algorithm in this paper has good performance and can accurately detect financial abnormal data.