Drug Audit Based on Bisecting K-Means Clustering Algorithm

To quickly discover suspicious data in massive data and facilitate the implementation of audit work, we need to use scientific experimental methods to conduct in-depth analysis of the collected data. This experiment uses bisecting k-means clustering algorithm which is an unsupervised learning method to realize in-depth analysis on the delivery data of 20 kinds of drugs from one hospital in 2018. During this experiment, three data features of drug delivery quantity, retail price and total price are selected as the clustering indexes, that is, the coordinate axes of the three-dimensional clustering diagram, and the optimal cluster number is determined according to SSE elbow rule. The experimental results show that the bisecting k-means algorithm can avoid falling into the local optimal state, it can be clearly seen that the suspicious phenomenon of some delivery drugs, so that auditors audit and verify the relevant departments and confirm the rectification. This experiment realized the drug audit based on bisecting k-means algorithm, and effectively solved the problem of multi-index visual analysis of complex data. Compared with the current audit, which uses audit software or database query statements to find suspicious clues, it innovates audit thinking and improves audit efficiency.