Keywords - Function groups - @ A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

Library: xclust
See also: cartsplit cartsplitopt cartcv leafnum maketr pred prederr prunecv pruneseq prunetot ssr kuva

Quantlet: prune
Description: Prunes a smaller regression tree from a tree which has been created by the cartsplit procedure. Uses error-complexity criterion.

Usage: subcs = prune (cs, alfa)
Input:
cs list of vectors: data structure which represents a binary tree and is produced by cartsplit procedure, contains vectors cs.val, cs.vec, cs.mean, cs.ssr, cs.nelem. See cartsplit for the description of cs.
alfa real number >= 0: the tree subcs is the subtree of cs which minimizes the error-complexity criterion with the complexity parameter alfa.
Output:
subcs similar kind of object than cs, see cartsplit for the description of subcs. The tree subcs is the subtree of cs which minimizes the error-complexity criterion when the complexity parameter is alfa.

Note:

Example:

; loads the library xclust

library ("xclust")

;let us generate a tree by cartsplit procedure

x1=#(0,0,0,0,1,1,1,1,1,2)

x2=#(0,0,0,0,0,0,0,1,1,1)

x=x1~x2

y=#(0,0,0,0,100,100,100,120,120,120)

cs=cartsplit(x,y,#(0,1))

cs

; let us choose a subtree when alfa=100

subcs=prune(cs,100)

subcs

Result:



Content of object cs.val.split0

[1,] 0 

[2,] 1,2 

Content of object cs.val.split1

[1,] NaN 

Content of object cs.val.split2

[1,] 0 

Content of object cs.val.split3

[1,] NaN 

Content of object cs.val.split4

[1,] NaN 

Content of object cs.vec

[1,] 1.000000 

[2,] NaN 

[3,] 2.000000 

[4,] NaN 

[5,] NaN 

Content of object cs.mean

[1,] 66.000000 

[2,] 0.000000 

[3,] 110.000000 

[4,] 100.000000 

[5,] 120.000000 

Content of object cs.var

[1,] 29640.000000 

[2,] 0.000000 

[3,] 600.000000 

[4,] 0.000000 

[5,] 0.000000 

Content of object cs.nelem

[1,] 10.000000 

[2,] 4.000000 

[3,] 6.000000 

[4,] 3.000000 

[5,] 3.000000 

Content of object subcs.val.split0

[1,] 0 

[2,] 1,2

Content of object subcs.val.split1

[1,] NaN 

Content of object subcs.val.split2

[1,] NaN 

Content of object subcs.vec

[1,] 1.000000 

[2,] NaN 

[3,] NaN 

Content of object subcs.mean

[1,] 66.000000 

[2,] 0.000000 

[3,] 110.000000 

Content of object subcs.var

[1,] 29640.000000 

[2,] 0.000000 

[3,] 600.000000 

Content of object subcs.nelem

[1,] 10.000000 

[2,] 4.000000 

[3,] 6.000000       


Library: xclust
See also: cartsplit cartsplitopt cartcv leafnum maketr pred prederr prunecv pruneseq prunetot ssr kuva

Keywords - Function groups - @ A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

Author: Jussi Klemela", 980323
(C) MD*TECH Method and Data Technologies, 21.9.2000