# [fpr 2006] \$BF|K\9TF07WNL3X2q%A%e!<%H%j%"%k%;%_%J!<(B

Kazunori Yamaguchi

```\$BN)65Bg3X\$N;38}\$G\$9!#(B

\$BF|K\9TF07WNL3X2q\$N%A%e!<%H%j%"%k%;%_%J!<\$,2<5-MWNN\$G3+:E\$5\$l\$^\$9!#(B
fpr\$B\$N%a%\$%j%s%0%j%9%H\$r\$*http://www.rikkyo.ne.jp/~kyamagu/BSJ/
\$B\$K>\:Y\$d?=\$79~\$_J}K!\$,MQ0U\$5\$l\$F\$\$\$^\$9\$N\$G(B
\$BJ;\$;\$F\$4Mw\$/\$@\$5\$\$!#(B

-----------------------------------------------------------------
\$BBjL\!'(B
Nonlinear Methods for Classification and Regression

\$B9V;U!'(B
\$B!!(BJerome H. Friedman\$B!J(BStanford\$BBg3X(B, \$B%"%a%j%+!K(B
\$B!!(BJacqueline J. Meulman\$B!J(BLeiden\$BBg3X(B, \$B%*%i%s%@!K(B

\$BFbMF!&\$M\$i\$\$!'(B

The purpose of the tutorial is to give an overview of state-of-the-art
nonlinear methods for the analysis of multivariate data. The latter would
typically describe the units of analysis (persons, objects, options,
consumers) on a large number of ordinal and nominal variables. The data
analysis approach advocated is particularly suited for recovery of
nonlinear relationships between the variables. The meteoric rise in
computing power has been accompanied by a rapid growth in the areas of
statistical modeling and multivariate data analysis. New techniques have
emerged for both predictive and descriptive learning that were not possible
in the past, using ideas that bridge the gap between statistics,
computer science and artificial intelligence. This tutorial will include
specific topics like decision trees, bagging and boosting, support vector
machines, nonlinear regression and classification with optimal scaling,
and the integration of the latter techniques with decision trees.
Attention will be given to the statistical aspects of their application,
and their integration with more standard statistical methodology.

\$BF|;~!'(B2001\$BG/(B9\$B7n(B13\$BF|(B13:00-17:00
\$B>l=j!'9C;R1`Bg3X(B

\$B;22CHqMQ(B(\$B;qNABe\$r4^\$`(B)\$B!'(B
\$B3X@8(B(\$B4^Hs2q0w3X@8(B) 2,000\$B1_(B, \$BF|K\9TF07WNL3X2q0w(B 4,000\$B1_(B, \$BHs2q0w(B 9,000\$B1_(B

\$BDj0w!'(B100\$BL>(B

\$B?=\$79~\$_!'%A%e!<%H%j%"%k%;%_%J!<\$N%5%\$%H(B
http://www.rikkyo.ne.jp/~kyamagu/BSJ/
\$B\$+\$i?=\$79~\$_\$/\$@\$5\$\$!#(B

\$B%*!<%,%J%\$%6!<(B
\$B!!N)65Bg3X!!http://www.ir.rikkyo.ac.jp/~kazunori/>
```

\$B%9%l%C%II=<((B \$BCx \$BF|IU=gI=<((B \$B%H%C%W%Z!<%8(B

\$B\$3\$3\$O(B\$B?4M}3X8&5f\$N4pAC%a!<%j%s%0%j%9%H(B\$B\$KEj9F\$5\$l\$?2a5n\$N5-;v\$r7G:\\$7\$F\$\$\$k%Z!<%8\$G\$9!#(B