基于NaiveBayes模型的贵州制造业绩效评价的分类研究

Research on the Classification of Manufacturing Performance Evaluation in Guizhou Based on NaiveBayes Model

摘要: 本文将NaiveBayes模型引入到贵州省制造业绩效评价的分类中，收集2014~2017年贵州省制造业529份经审计无保留意见财务报表数据，结合国家年度公布的《国家企业绩效评价标准值》，分别对盈利能力、营运能力、偿债能力三个方面财务能力下的绩效因子进行分段量化。通过财务报表科目构造一系列的变量，建立企业绩效评价的财务能力分类模型，探索制造业的绩效评价内容与财务报表其他科目之间的关系。在Accuracy和AUC评价准则下，通过模型训练和实证分析结果表明NaiveBayes模型的表现优于logistic回归模型，BP神经网络，二分类SVM以及决策树。

Abstract: In this paper, NaiveBayes model is introduced into the classification of performance evaluation of manufacturing industry in Guizhou Province. 529 audited and unqualified financial statement data of manufacturing industry in Guizhou Province from 2014 to 2017 are collected. Combined with the standard value of enterprise performance evaluation published by the state, the performance factors of profitability, operation ability and solvency are quantified by sections. By constructing a series of variables, this paper establishes the financial ability classification model of enterprise performance evaluation, and explores the relationship between the performance evaluation content of manufacturing industry and other subjects of financial statements. Under the evaluation criteria of accuracy and AUC, the results of model training and empirical analysis show that the performance of NaiveBayes model is better than that of logistic regression model, BP neural network and binary SVM and decision tree.