BBT rating algorithm Wrapper arround `bbt` update algorithm. Wrapper allows user to simplify calculation providing only data and initial parameters assumptions
bbt_run(formula, data, r, rd, sigma, weight, beta = 25/6, kappa = 0.5,
gamma, idlab, init_r = 25, init_rd = 25/3, pb = FALSE)formula specifying model. BBT algorithm allows only player ranking parameter and should be specified by following manner: `rank | id ~ name`. Names in formula are unrestricted, but model structure remains the same:
rank player position in event.
id event identifier in which pairwise comparison is assessed.
name of player.
data.frame which contains columns specified in formula, and optionaly columns defined by `sigma`, `weight` or `date`.
named vector of initial rating estimates. In there is no assumption, initial ratings is set to be r=25 Names of vector should correspond with `name` in formula.
named vector of initial rating deviation estimates. In there is no assumption, initial is set to be r=25/3 Names of vector should correspond with `name` in formula.
name of column in `data` containing rating volatility. Rating volitality is a value which multiplies prior `rd`. If `sigma > 0` then prior `rd` increases, making estimate of `r` more uncertain.
name of column in `data` containing weights. Weights increasing or decreasing update change. Higher weight increasing impact of corresponding event.
The additional variance of performance. As beta increases, the performance is more uncertain and update change is smaller. By default `beta = 25/6`.
controls `rd` shrinkage not to be greater than `rd*(1-kappa)`. `kappa=1` means that `rd` will not be decreased.
can help to control how fast the variance `rd` is reduced after updating. Lower `gamma` slows decreasing of `rd`, which tends to reach zero to quickly. The default value is `gamma = rd/c`.
name of column in `data` containing date. Doesn't affect estimation process. If specified, charts displays estimates changes in time instead of by observation `id`
initial values for `r` if not provided. Default = 25
initial values for `r` if not provided. Default = 25/3
logical, if TRUE progress bar will appear in console. Default = FALSE
A "ratings" object is returned
final_r named vector containing players ratings.
final_rd named vector containing players ratings deviations.
r data.frame with evolution of the ratings and ratings deviations estimated at each event.
pairs pairwise combinations of players in analysed events with prior probability and result of a challange.
class of the object
method type of algorithm used
formula modelled formula
# NOT RUN {
# Example from Glickman
data <- data.frame( name = c( "A", "B", "C", "D" ),
rank = c( 3, 4, 1, 2 ))
bbt <- bbt_run( rank ~ name, data )
# }
Run the code above in your browser using DataLab