Dropping repeated columns from a matrix

[Shoummo Sen Gupta]

Dear statalist,

I have a matrix in which two or more columns are exactly same. Is there some way of keeping one of them and dropping all others?

For example, I have the matrix

1 4 2 1 1
2 7 0 2 2
3 3 5 3 3

I want to drop the last two columns since they are same as the 1st column.

Thank you in advance.

Shoummo Sen Gupta

[Clyde Schechter]

Mata seems to have a command or function for almost any imaginable operation on matrices, but I don't know one for this. That's probably just one of many gaps in my knowledge of Mata, and perhaps somebody else will show you a simpler solution. But this will do it:

clear*

matrix M = (    1, 4, 2, 1, 1 \ ///
                2, 7, 0, 2, 2 \ ///
                3, 3, 5, 3, 3)
               
matrix list M

matrix MT = M'

svmat MT
duplicates drop

mkmat MT*, matrix(MT)
matrix M = MT'

matrix list M

[daniel klein]

Mata has a uniqrows() function for that. Obviously, you need to transpose the matrix before using the function. A side-effect is that the matrix is sorted. My elabel (SSC) package has a Mata function, distinctrowsof(), that preserves the original sort order.

Best
Daniel

[Shoummo Sen Gupta]

Thanks all.. These will help.