摘要
"大数据战略"给税务数据分析工作带来了新的机遇。基于中国S省房地产业的税务登记、申报征收、财务报表等涉税数据,利用计算机学习建立纳税遵从风险模型,识别出风险等级,运用关联规则算法将影响纳税遵从风险的因子精确到收入、成本和费用等财务指标上。研究表明,"机器学习+关联规则"组合算法,支持规模超大、关系错综复杂的数据信息,可以有效识别纳税遵从风险等级并深度挖掘造成不同纳税遵从风险的原因,该方法的普适性对微观数据、复杂数据和大数据的分析具有参考价值。
The strategy of ＂big data＂ has brought new opportunities to the tax data analysis work, we have studied the real estate industry in S Province of China. By establishing a tax compli- ance risk model using the machine learning method, we have identified the risk level of taxpayers. We then set up a quantitative analysis model, which includes the latest technology and combines the taxation and economy reality of China, by using association rule algorithm and expressing enterpri- ses＂ tax compliance risks as financial indicators such as incomes, costs and expenses. Results show that the combination ＂machine learning ＋ association rules＂ algorithm, support the large scale and complex relationship between the collection of data and information, can effectively identify tax compliance risk grade and the depth of mining the cause of different tax compliance risk. The com- bined use of the above algorithm has the reference value for micro data, complex data, data analy- sis.
出处
《现代财经（天津财经大学学报）》
CSSCI
北大核心
2015年第11期46-59,共14页
Modern Finance and Economics：Journal of Tianjin University of Finance and Economics
基金
国家自然科学基金项目(71373267)
关键词
大数据
风险识别
机器学习
关联规则
Big Data
Risk Identification
Machine Learning
Association Rules
