Sort or rank the rows of a matrix

tda_msort sorts the rows of X ascending by the columns named in by, first column first – TDA's msort, matching X[do.call(order, as.data.frame(X[, by])), ]. tda_msort1 does the same and then drops consecutive duplicate rows (comparing every column, not just by) – TDA's msort1, matching unique(sorted_X). tda_mrank – despite the name – is not a linear-algebra matrix rank (no TDA command returns one as a matrix value; tda_mginv prints a pseudorank as a side effect, but never returns it). It returns the sort permutation itself: an integer vector p such that X[p, ] is sorted by by – TDA's mrank, which equals base R's order(X[, by[1]], X[, by[2]], ...), not rank().

Usage

tda_msort(X, by, dir = tempfile("tda"))

tda_msort1(X, by, dir = tempfile("tda"))

tda_mrank(X, by, dir = tempfile("tda"))

Arguments

X

a numeric matrix.

by

which columns to sort by, in order (1-based, at least one).

dir

working directory.

Value

tda_msort/tda_msort1 return a matrix; tda_mrank returns an integer vector, the same length as nrow(X).

See also

Other matrix reshaping: tda_mag(), tda_mcath(), tda_mcvec(), tda_mrsum(), tda_msrow(), tda_mtrim()

Examples

X <- rbind(c(30, 1), c(10, 1), c(20, 1), c(40, 1))
tda_msort(X, 1)
#>      [,1] [,2]
#> [1,]   10    1
#> [2,]   20    1
#> [3,]   30    1
#> [4,]   40    1
tda_mrank(X, 1)   # == order(X[, 1]), i.e. c(2, 3, 1, 4)
#> [1] 2 3 1 4