改进退火遗传算法在SQL数据挖掘参数优化中的应用

Application of Improved Simulated Annealing Genetic Algorithm in SQL Data Mining Parameter Optimization

摘要:

针对SQL数据挖掘在复杂动力学系统故障诊断中的模式分类问题,以决策树参数优化为例,开展SQL数据挖掘分类算法参数优化研究。目前数据挖掘中的各类算法参数往往根据经验值设定,预测精度不高；只用遗传算法进行参数优化,分类预测结果容易发生振荡和早熟现象。采用改进的退火遗传算法对SQL数据挖掘中的决策树算法参数进行优化,解决了人工经验设置参数效率低下、精度不高的问题,同时实现了全局搜索,快速收敛到全局最优解。

Abstract:

This research focused on the classification of SQL server data mining for fault diagnosis of complex dynamic system model, and carried out the optimization of parameter on SQL data mining classification algorithm, which taking parameter optimization decision tree as an example. There is a problem in data mining that the parameters settings are chosen by manual and the accuracy of prediction is unsatisfactory. In addition, parameter optimization with simple GA algorithm is prone to oscillation and prematurity. For low accuracy of artificial parameters settings and unsatisfactory results, an improved annealing genetic algorithm in SQL server data mining has been proposed to optimize the parameter of decision tree. Moreover, what have been achieved are that the global search and fast convergence to the global optimal solution.