EloOptimized provides tools to implement the maximum likelihood methods for deriving Elo scores as published in Foerster, Franz et al. (2016). Chimpanzee females queue but males compete for social status. In addition, it provides functionality to efficiently generate traditional Elo scores using a simplified procedure that doesn’t require the use of cumbersome presence matrices. Finally, it quickly generates a number of additional Elo-based indices (ordinal, normalized, cardinal, and categorical ranks and rank scores) of potential use to researchers, as outlined in the linked manuscript.
# Current version on Github:
# install.packages("devtools")
devtools::install_github("jtfeld/EloOptimized")
# CRAN-approved version on CRAN:
install.packages("EloOptimized")
There are two functions of interest. Use eloratingopt() to calculate Elo scores using optimized Elo parameter values, or eloratingfixed() to calculate Elo scores using user-defined parameter values.
# to generate Elo scores using fixed initial Elo scores (1000) and a ML-fitted value for the K parameter:
nbaelo = eloratingopt(agon_data = nba, fit_init_elo = FALSE)
# to generate Elo scores using fixed default initial Elo scores and default K:
nbaelo = eloratingfixed(agon_data = nba, k = 100, init_elo = 1000)
To recreate the results from the 2016 manuscript, use the following code:
# Males, model type 1:
melo1 = eloratingopt(agon_data = chimpagg_m, pres_data = chimppres_m, fit_init_elo = F)
# Males, model type 3:
melo3 = eloratingopt(agon_data = chimpagg_m[101:nrow(chimpagg_m),],
pres_data = chimppres_m, fit_init_elo = T)
# Females, model type 1:
felo1 = eloratingopt(agon_data = chimpagg_f, pres_data = chimppres_f, fit_init_elo = F)
# Females, model type 3:
felo3 = eloratingopt(agon_data = chimpagg_f[101:nrow(chimpagg_f),],
pres_data = chimppres_f, fit_init_elo = T)