Design of an Early Warning Intelligent Model for Enterprise Industrial and Commercial Crisis based on Multi-Dimensional Computing Data Outlier Analysis

Abstract

This paper analyzes and studies the effect of the selection of the number of real clusters on the outlier detection in the cluster-based outlier detection algorithm and proposes an outlier detection algorithm based on the automatic clustering method. Firstly, factor analysis, mean test and correlation analysis are used to screen financial indicators and corporate governance variables, respectively, to obtain representative indicator variables, and then use support vector machine method to conduct empirical analysis. The research results show that the support vector machine model has a strong predictive ability for enterprise bankruptcy risk. Using the distance relationship between each candidate outlier and its data block and adjacent data blocks, it is determined which processor each candidate outlier needs to carry out network communication. Metrics, Alerts, Diagnoses and Gaining Experience.