zapweak: Zap weak competitors

In hyper2: The Hyperdirichlet Distribution, Mark 2

Zap weak competitors

Description

Given a hyper2 object, discard competitors with a small estimated strength.

Usage

zapweak(H, minstrength = 1e-05, maxit, ...)

Arguments

H	Object of class hyper2
minstrength	Strength below which to discard competitors
maxit	Maximum number of iterations; if missing, use size(H)-1
...	Further arguments, passed to maxp()

Details

Iteratively discards the weakest player (if the estimated strength is less than minstrength) using discard_flawed(). maxp(..,n=1) for efficiency.

Value

Returns a slimmed-down hyper2 object with weak players removed.

Note

This function is experimental and appears to be overly aggressive. For some likelihood functions zapweak() removes all the players.

I now think that there is no consistent way to remove weaker players from a likelihood function. I think the only way to do it is to look at the dataset that generates the likelihood function, somehow weed out the players with the poorest performance, and generate a new likelihood function without them.

Author(s)

Robin K. S. Hankin

See Also

discard_flawed,maxp

Examples

 zapweak(icons)        # removes noone
# zapweak(rowing)      # removes everyone...

hyper2 documentation built on June 22, 2024, 9:57 a.m.