摘要
提出了一种多元数据的点得分平行坐标表示及可视化分析方法。该方法利用简单贝叶斯公式计算各属性值或属性值区间的频数和点得分,最后根据构建的点得分平行坐标即可进行数据集的可视化分析和未知样本的分类。将该方法应用到一个肝功异常数据集的结果表明,利用该图表示可以有力地揭示数据内在结构和发现知识,从而特别适合应用到疾病诊断等数据分析领域。
A new multivariate data visualization method named point-score parallel coordinates （PSPC） is proposed in this paper. This method computes the frequency numbers and point scores through the naive Bayes formula, then the dataset can be analyzed and classified visually by the constructed PSPC. The result of this method applied to a liver disorders dataset indicates that it has the merits of exposing the structure of original data set and facilitating the discovery of knowledge, consequently it is very fit to be applied in many data analysis domains such as medical diagnostics.
出处
《燕山大学学报》
CAS
2008年第5期440-444,470,共6页
Journal of Yanshan University
基金
国家自然科学基金资助项目(60605006)
关键词
多元数据
平行坐标
简单贝叶斯
点得分
multivariate data
parallel coordinates
naive Bayes
point scores
