[R] obtaining the discriminant line from lda
Clayton Springer
csprin at brandybuck.ca.sandia.gov
Fri Apr 28 22:15:50 CEST 2000
Dear R folks,
Thanks to all your help before I have loaded a 1-D toy data set into
R and did LDA on it. The toy data has Class=0 if value>0.
> XY <-- read.table ("test.xy",header=T )
> XY
X.Class value
1 0 60.4897262
2 0 32.9554489
3 -1 -53.6459189
4 0 44.4450579
.
.
.
998 -1 -43.4183157
999 0 7.9865092
1000 -1 -8.2279180
> XY.lda <- lda(X.Class ~ value,XY)
> XY.lda
Call:
lda.formula(X.Class ~ value, data = XY)
Prior probabilities of groups:
-1 0
0.521 0.479
Group means:
value
-1 -48.66322
0 49.91819
Coefficients of linear discriminants:
LD1
value 0.0357248
> XY.lda$svd
[1] 55.63543
> XY.lda$class
NULL
> XY.lda$posterior
NULL
Question #1: How do I obtain the line that lda thinks divides the
two groups? (which here it is between 1 and 2.)
Next I load in a test set for prediction:
> Predict0
value
1 -10
2 -9
3 -8
4 -7
5 -6
6 -5
7 -4
8 -3
9 -2
10 -1
11 0
12 1
13 2
14 3
15 4
16 5
17 6
18 7
19 8
20 9
21 10
> Predict0.lda <- predict(XY.lda,Predict0)
> Predict0.lda$class
[1] -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 0 0 0 0 0 0 0 0 0
For those who don't want to count this shows that the dividing
line is somewhere between 1 & 2, even though my toy data set
can be perfectly divided at 0. I had not expected (Fischer's) LDA
to behave this way.
Question #2: Are there parameter adjustments and/or other LDA methods
where I can get the expected dividing surface at 0. (presumability
a classification tree would choose the line I desire, but I want
a lda method that does this.)
If I am not being clear let me know,
Thanks in advance,
Clayton
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
More information about the R-help
mailing list