几何尺寸与公差论坛------致力于产品几何量公差标准GD&T (GDT:ASME)|New GPS(ISO)研究/CAD设计/CAM加工/CMM测量 - 查看单个帖子 - statistics of k-fold decision tree using jmp

statistics of k-fold decision tree using jmp
hi everyone,
i'm looking for some help to interpret a decision tree result using k-fold cross validation in jmp statistical software.
basically, after partionning my data where both response and variables are continuous, i got a tree with the decision rules and a statistics associated with the k-fold cross validation. are displayed, the number of cross validation rounds (let say k-fold=10) then the two following statistics :
(i) crossval sse = 29.96
(ii) sse = 21.89
how can i interpret this ? i understand from help in the software that because i have continuous response, sum of squares are calculated.
many thanks in advance for kind suggestions
paleolithic
cross validation is a type of test in which a data set is split in "k" sub-groups. the model is tested on those sub-groups that were not used for the model building. so your cross-validation sse (sum of squared errors) is very close to the initial model sse, meaning it is robust.
thumbs up ! thanks mr mask

2009-09-05, 09:05 PM	#1
huangyhg 超级版主注册日期: 04-03 帖子: 18592 精华: 36 现金: 249466 标准币资产: 1080358888 标准币	statistics of k-fold decision tree using jmp statistics of k-fold decision tree using jmp hi everyone, i'm looking for some help to interpret a decision tree result using k-fold cross validation in jmp statistical software. basically, after partionning my data where both response and variables are continuous, i got a tree with the decision rules and a statistics associated with the k-fold cross validation. are displayed, the number of cross validation rounds (let say k-fold=10) then the two following statistics : (i) crossval sse = 29.96 (ii) sse = 21.89 how can i interpret this ? i understand from help in the software that because i have continuous response, sum of squares are calculated. many thanks in advance for kind suggestions paleolithic cross validation is a type of test in which a data set is split in "k" sub-groups. the model is tested on those sub-groups that were not used for the model building. so your cross-validation sse (sum of squared errors) is very close to the initial model sse, meaning it is robust. thumbs up ! thanks mr mask __________________ 借用达朗贝尔的名言：前进吧，你会得到信心! [url="http://www.dimcax.com"]几何尺寸与公差标准[/url]