首页> 外国专利> METHOD OF DISCRETING / GROUPING A SOURCE ATTRIBUTE OR A GROUP ATTRIBUTES SOURCE OF A DATABASE

METHOD OF DISCRETING / GROUPING A SOURCE ATTRIBUTE OR A GROUP ATTRIBUTES SOURCE OF A DATABASE

机译:数据库的源属性或组属性的离散/分组方法

摘要

The present invention relates to a method for discretizing / grouping a source attribute or a source attribute group of a database containing a population of individuals, in particular for the purpose of predicting modalities of a target attribute. given. Said method comprises the following steps of: a) partitioning said modalities of said source attribute or said attribute group into elementary regions, b) Evaluation for each pair of elementary regions, of a merge criterion, c) Search, among the together couples of mergeable elementary regions, the pair of elementary regions whose fusion criterion would be optimized, e) Stopping the method if there are no elementary regions whose fusion would have the consequence of improving said merger criterion f) otherwise merging and repeating steps b) to e) .In accordance with the present invention, it further comprises a step d) between steps c) and e) direct jump to step f) as long as the value d an appreciation variable of the merger considered is not included in a predetermined atypical value zone. / P
机译:本发明涉及一种用于对包含一群个体的数据库的源属性或源属性组进行离散化/分组的方法,特别是为了预测目标属性的模态。给定的。所述方法包括以下步骤:a)将所述源属性或所述属性组的所述模态划分为基本区域; b)对每对基本区域的评估,合并准则; c)在可合并的夫妇之间进行搜索。基本区域,一对其融合标准将被优化的基本区域,e)如果没有任何融合将改善所述合并标准的结果的基本区域,则停止该方法f)否则合并并重复步骤b)至e)。根据本发明,只要在所考虑的合并的增值变量的值d不包括在预定的非典型值区域中,则在步骤c)和e)之间还包括直接跳到步骤f)的步骤d)。

著录项

  • 公开/公告号FR2849249A1

    专利类型

  • 公开/公告日2004-06-25

    原文格式PDF

  • 申请/专利权人 FRANCE TELECOM;

    申请/专利号FR20020016733

  • 发明设计人 MARC BOULLE;

    申请日2002-12-19

  • 分类号G06N5/02;G06F17/10;

  • 国家 FR

  • 入库时间 2022-08-21 22:39:16

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号