基于一种优化权重的K-近邻算法的缺失财务数据组合填补

K-nearest Neighbor Algorithm Based on Optimized Weight to Fill in the Missing Financial Data Combination

在对上市公司进行财务分析时,相关财务数据的完整性对数据分析意义重大.针对这些非随机性缺失财务数据的填补,将基于明考夫斯基距离,根据近邻原理找出与缺失数据同类的相似样本,作为一个新的训练数据集;然后提出一种优化权重的K-近邻算法(OKNN算法),对上述的相似样本给予各个指标组合权重系数值;最后根据优化的权重系数对提出的新算法进行实例验证,结果证实提出的OKNN组合填补方法优于经典的KNN算法和加权的KNN算法距离填补法.

When conducting financial analysis of listed companies,the completeness of relevant financial data is of great significance to data analysis.For the filling of these non-random missing financial data,samples similar to the missing data were found based on the Minkowski distance and the nearest neighbor principle as a new training data set.Then,a K-nearest neighbor algorithm(OKNN)with optimized weight was proposed,and the combined weight of each index were given to these similar samples.Finally,an example was given to verify the proposed new algorithm according to the optimized weight coefficient.The results showed that the proposed OKNN combined filling method was superior to the classical KNN algorithm and the weighted KNN algorithm distance filling method.