[R] Could not fit correct values in discriminant analysis by bruto.

[Kuhn, Max]

Shuji,

I suspect that bruto blows up because your data are linearly separable.
To see this (if you didn't already know), try

   library(lattice)
   splom(~x, groups = y)

and look at the first row. If you are trying to do classification, there
are a few methods that would choke on this (logistic regression) and a
few that won't (trees, svms etc). I would guess that bruto is in the
latter group.

However, if you are try to do classification, try using bruto via fda:

   > tmp <- cbind(x, factor(y))
   > 
   > fdaFit <- fda(y2~., tmp)
   > fdaFit
   Call:
   fda(formula = y2 ~ ., data = tmp)
   
   Dimension: 1 
   
   Percent Between-Group Variance Explained:
    v1 
   100 
   
   Degrees of Freedom (per dimension): 5 
   
   Training Misclassification Error: 0 ( N = 20 )
   > 
   > predict(fdaFit, type = "posterior")[1:3,]
     0 1
   2 0 1
   2 0 1
   2 0 1

Max

-----Original Message-----
From: r-help-bounces at stat.math.ethz.ch
[mailto:r-help-bounces at stat.math.ethz.ch] On Behalf Of ?? ??
Sent: Sunday, April 08, 2007 10:47 PM
To: r-help at stat.math.ethz.ch
Subject: [R] Could not fit correct values in discriminant analysis by
bruto.

Dear R-users,

I would like to use "bruto" function in mda package for flexible  
discriminant analysis.
Then, I tried, for example, following approach.

 > x
             band1              band2            band3
1   -1.206780      -1.448007      -1.084431
2   -0.294938      -0.113222      -0.888895
3   -0.267303      -0.241567      -1.040979
4   -1.206780      -1.448007      -1.040979
5   -1.151518      -0.806286      -0.671630
6   -1.179146      -1.396670      -1.453775
7   -0.294938      -0.241567      -1.453775
8   -0.350200      -0.267239      -1.084431
9   -1.151518      -0.857623      -0.649901
10  1.362954      -1.396670      -2.235926
11 -0.239675       1.118883       1.457551
12 -0.294938      -1.268325      -0.497817
13 -0.294938      -0.729278      -0.106745
14 -1.123883      -0.703612      -0.150196
15  0.616905       1.144548       -0.150196
16 -0.267303      1.657930         1.044750
17  1.611637      1.041874          0.610225
18 -1.123883     -0.677941         0.262605
19 -0.239675     -0.626604        -0.128473
20  2.274797       1.118883         1.805171

 > y
[1] 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0

 > fit <- bruto(x,y)

But, obtained fit$fitted.values are enormously high (or low) .
Execution of bruto(x[,2:3], y) is done well (values are nearly 1 or 0).
Values of column 1 are wrong or appropriate option is needed?
I contacted the package maintainer, but the problem could not be solved.

Thanks

Shuji Kawaguchi

 > R.version
platform       i386-apple-darwin8.8.1
arch           i386
os             darwin8.8.1
system         i386, darwin8.8.1
version.string R version 2.4.0 (2006-10-03)

______________________________________________
R-help at stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

----------------------------------------------------------------------
LEGAL NOTICE\ Unless expressly stated otherwise, this messag...{{dropped}}