convert r code into mata

[Mike Lauder]

Dear list members,

I hope somebody can help me with this problem. I'd like to run a small R code in stata and mata. The existing R code (see below) calculates a distance matrix (cosine distance) based on variables x y z in a dataset (testdata.csv) and writes the distance matrix in a new file (distancematrix_test.csv).

This is the R program, including some output based on a test dataset:

> # load data
> mydata = read.csv("testdata.csv")

> # View the data
> mydata
x y z
1 0.5 0 0.5
2 0.5 0 0.5
3 0.5 0 0.5
4 0.5 0 0.0
5 1.0 0 0.0

> # define distance function (cosine distance)
> # and perform function on mydata (this should be done in MATA)

> cosineDistance <- function(x){
as.dist(1 - x%*%t(x)/(sqrt(rowSums(x^2) %*% t(rowSums(x^2)))))
}

> # perform distance function on mydata
> s <- data.matrix(mydata)

> # view result (distance matrix)
> cosineDistance(s)
1 2 3 4
2 0.0000000
3 0.0000000 0.0000000
4 0.2928932 0.2928932 0.2928932
5 0.2928932 0.2928932 0.2928932 0.0000000

> # write result (distance matrix) to disk
> x <- cosineDistance(s)
> m <- as.matrix(x)
> write.table(m, file = "distancematrix_test.csv")


Any help is much appreciated!

Thanks a lot!!

Mike

[Hua Peng (StataCorp)]

Try the following:

cscript
mata:
real matrix cosineDistance(real matrix X)
{
    real rowvector norm
    real matrix res, rr
    
    /* compute norm of X for each row vector*/
    norm = sqrt(rowsum(X:^2))'
    rr = cross(norm, norm)
    res = 1:-cross(X', X'):/rr
    return(res)
}
end

/* read the dataset from csv */
import delimited testdata.csv, varname(1)

/* put dataset into a Mata matrix */
putmata X=(x y z)

mata:
X
D=cosineDistance(X)
D
end

/* put the distance matrix back into a Stata dataset */
drop _all
getmata (d*)=D

/* save the datset to a csv file */
export delimited distncematrix_test.csv, replace

[Mike Lauder]

It works!! Thank you so much -- you saved my life!!!

Mike