基于概率统计的小差异数据的分类模型仿真

The Small Difference Data Classification Model Based on Probability and Statistics Simulation

小差异数据往往具有信息量大、特征差异小的特点,传统的数据分类方法多具有串行性,在处理海量小差异数据时,存在分类效率低、准确率低及可行性差的缺陷,为信息检索、数据管理等实际应用带来了潜在的风险。为此,提出设计一种基于概率统计的小差异数据分类模型。针对初始数据的杂乱性、冗余性和随机性,分别进行数据清洗、数据变换和数据归约等预处理,依据相关原理构建隐马尔科夫数据分类模型,并通过模型参数优化,得到数据特征的最优描述及该数据属于每一类别的最大概率值,从而实现小差异数据的有效分类。实验结果表明,采用改进算法进行小差异数据分类,能够大大提高数据分类的准确性,提升系统运行速率,提高了算法鲁棒性,具有实际的应用价值。

Small difference data tend to have large amount of information, the difference of characteristics of the characteristics of small, traditional methods of data classification have serial more, in dealing with massive data, small differences exist low classification efficiency, low accuracy and feasibility of defects, for information retrieval, data management, and other practical application has brought the potential risks. For this, put forward to design a small difference data classification model based on probability and statistics. On the initial data of , redundancy and randomness, respectively for data cleaning, data transformation and data reduction, such as pretreatment, hidden markov data classification model was constructed according to the relevant principles, and through the model parameter optimization, the optimal characteristics of the data description is obtained and the data belong to each other's largest probability value, so as to realize the effective classification of small differences in data. Experimental results show that the improved algorithm is adopted to improve the small difference data classification, can greatly improve the accuracy of data classification, improve the system operation speed, improved the algorithm robustness, has a great application value.