ISI {EloRating}

de Vries' I&SI ranking

Description

de Vries' I&SI ranking

Usage

ISI(mat, runs = 5000, printmessages = TRUE)

Arguments

mat

square interaction matrix with winner in rows and losers in columns, for example the output from creatematrix

runs

numeric, number of iterations, by default 5000

printmessages

logical, should the number of I and SI be printed (as well as a message if there is more than one solution). By default TRUE.

Details

The number of interations is set substantially higher than what was suggested in the de Vries' 1998 paper, because my algorithm here is less efficient.

The I&SI algorithm (c.f. de Vries 1998) does not necessarily result in a unique order (see example below). If such a case occurs, all (equally good) solutions are returned as a list.

The function checks whether a table is supplied instead of a matrix and converts from table to matrix if possible (trying to keep the column and row names if supplied in the table).

If the matrix does not have column-names, unique column- and row-names are assigned.

Value

a list with the best possible matrix (or matrices if there is more than one best solution)

Author(s)

Christof Neumann

References

de Vries H (1998). “Finding a dominance order most consistent with a linear hierarchy: a new procedure and review.” Animal Behaviour, 55, 827-843. doi:10.1006/anbe.1997.0708.

See Also

ISIranks

Examples

 data(devries98)
 h.index(devries98)
 ISI(devries98)

 ##
 data(adv)
 SEQ <- elo.seq(winner=adv$winner, loser=adv$loser, Date=adv$Date)
 mat <- creatematrix(SEQ)
 res <- ISI(mat)
 # note that this matrix is not sufficiently linear to justify such ordering
 h.index(mat)