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 prune prunecv pruneseq prunetot kuva

Quantlet: ssr
Description: Returns the sum of squared residuals in a regression tree.

Usage: res = ssr(cs, node)
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.
node integer >= 1: gives the index of the root node of the subtree whose mean of squared residuals will be calculated. If node=1, the mean of squared residuals in the hole tree is calculated, if node=2, the mean of squared residuals in the left subtree is calculated, if node=3, the mean of squared residuals in the left subtree of the left subtree (if it exists) is calculated. See cartsplit for the explanation how the tree is represented as a vector.
Output:
res real number >= 0: sum of squared residuals in the specified subtree.

Note:

Example:

; load 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

val1=ssr(cs,1)

val1

prcs=prunetot(cs,2)

val2=ssr(prcs,1)

val2

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.ssr

[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 val1

[1,] 0.000000 

Content of object prcs.val.split0

[1,] 0 

[2,] 1,2 

Content of object prcs.val.split1

[1,] NaN 

Content of object prcs.val.split2

[1,] NaN 

Content of object prcs.vec

[1,] 1.000000 

[2,] NaN 

[3,] NaN 

Content of object prcs.mean

[1,] 66.000000 

[2,] 0.000000 

[3,] 110.000000 

Content of object prcs.ssr

[1,] 29640.000000 

[2,] 0.000000 

[3,] 600.000000 

Content of object prcs.nelem

[1,] 10.000000 

[2,] 4.000000 

[3,] 6.000000 

Content of object val2

[1,] 600.000000   


Library: xclust
See also: cartsplit cartsplitopt cartcv leafnum maketr pred prederr prune prunecv pruneseq prunetot 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 Klemelae, 980323
(C) MD*TECH Method and Data Technologies, 21.9.2000