R help - Aug 2012 - Different results between lda(mass) and spss discriminant analysis

Hi all,
I obtained a strage result with LDA (MASS) function in R with NIR data.
I tried both CV (leave one out cross validation) and splitting my data in
odd (training) and even (prediction) sets.
In all the cases the minimum error was near to 0.

Due to the strange result, I tried with SPSS IBM software and it give me
around 11% of minimum error with and without leave one out cross validation.

Maybe the problem is a my error in my script?

Someone can check it pls?

data <- "data\\raw_data.csv"

r <- read.csv(data, header = T)
sound <- r[1:844,]
unsound <- r[845:2195,]

even_s <- seq(nrow(sound)) %% 2
even_u <- seq(nrow(unsound)) %% 2 
t <- rbind(sound[even_s == 1,], unsound[even_u == 1,])
p <- rbind(sound[even_s != 1,], unsound[even_u != 1,])

fit <- lda(samples ~., data = t)
ct <- table(p[, 1], predict(fit, p[,-1])$class)
errors <- 1-diag(prop.table(ct, 1))
min.err <- 1-sum(diag(prop.table(ct)))

fit_cv <- lda(samples ~., data = r, CV =T)
ct_cv <- table(r[,1], fit_cv$class)
errors_cv <- 1-diag(prop.table(ct_cv, 1))
min.err_cv <- 1-sum(diag(prop.table(ct_cv)))

thank you for your help!

Best regards,
Roberto



--
View this message in context:
http://r.789695.n4.nabble.com/Different-results-between-lda-mass-and-spss-discriminant-analysis-tp4638763.html
Sent from the R help mailing list archive at Nabble.com.

I don't see anything obviously wrong, but I seem to remember that SPSS uses
equal prior probabilities as the default whereas lda() uses "the class
proportions for the training set." That might explain the difference.

----------------------------------------------
David L Carlson
Associate Professor of Anthropology
Texas A&M University
College Station, TX 77843-4352

> -----Original Message-----
> From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-
> project.org] On Behalf Of Roberto
> Sent: Wednesday, August 01, 2012 5:20 PM
> To: r-help at r-project.org
> Subject: [R] Different results between lda(mass) and spss discriminant
> analysis
> 
> Hi all,
> I obtained a strage result with LDA (MASS) function in R with NIR data.
> I tried both CV (leave one out cross validation) and splitting my data
> in
> odd (training) and even (prediction) sets.
> In all the cases the minimum error was near to 0.
> 
> Due to the strange result, I tried with SPSS IBM software and it give
> me
> around 11% of minimum error with and without leave one out cross
> validation.
> 
> Maybe the problem is a my error in my script?
> 
> Someone can check it pls?
> 
> data <- "data\\raw_data.csv"
> 
> r <- read.csv(data, header = T)
> sound <- r[1:844,]
> unsound <- r[845:2195,]
> 
> even_s <- seq(nrow(sound)) %% 2
> even_u <- seq(nrow(unsound)) %% 2
> t <- rbind(sound[even_s == 1,], unsound[even_u == 1,])
> p <- rbind(sound[even_s != 1,], unsound[even_u != 1,])
> 
> fit <- lda(samples ~., data = t)
> ct <- table(p[, 1], predict(fit, p[,-1])$class)
> errors <- 1-diag(prop.table(ct, 1))
> min.err <- 1-sum(diag(prop.table(ct)))
> 
> fit_cv <- lda(samples ~., data = r, CV =T)
> ct_cv <- table(r[,1], fit_cv$class)
> errors_cv <- 1-diag(prop.table(ct_cv, 1))
> min.err_cv <- 1-sum(diag(prop.table(ct_cv)))
> 
> thank you for your help!
> 
> Best regards,
> Roberto
> 
> 
> 
> --
> View this message in context: http://r.789695.n4.nabble.com/Different-
> results-between-lda-mass-and-spss-discriminant-analysis-tp4638763.html
> Sent from the R help mailing list archive at Nabble.com.
> 
> ______________________________________________
> R-help at r-project.org 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.