data-raw/temp.R

In elo2zero/sport: Sequential Pairwise Online Rating Techniques

library(data.table)
library(psych)
library(sport)
df <- data.table(
  team = c("A", "B", "C", "D"),
  rank = c(3, 4, 1, 2),
  days = c(0, 0, 0, 0),
  r = c(1500, 1400, 1550, 1700),
  rd = c(200, 30, 100, 300),
  sig = c(.06, .06, .05, .07),
  tau = .5
)

mi_ij <- matrix(c(1500, 1400, 1550, 1700), 2, 2)
sig_ij <- matrix(c(200, 30, 100, 300), 2, 2)
sig2_ij <- sig_ij^2
rank_i <- c(1, 2)


mi_ij <- matrix(c(1500, 1400, 1550, 2000), 2)
sig_ij <- matrix(c(200, 30, 100, 200), 2)
rank_i <- c(1, 2)
kappa <- 0.95

set.seed(1)
ddl(
  teams = c("A", "B", "C", "D"),
  rank = c(3, 4, 1, 2),
  X = matrix(rnorm(4), ncol = 1),
  H = matrix(1, nrow = 4),
  S = matrix(rlnorm(4), nrow = 4)
)

set.seed(1)
ddl(
  teams = c("A", "B", "C", "D"),
  rank = c(3, 4, 1, 2),
  X = matrix(c(1.2, -0.5, 0.1, 0.9), ncol = 1),
  H = matrix(c(1, 1, 1, 1), ncol = 1),
  S = matrix(rlnorm(4), ncol = 1)
)

ddl(
  teams = c("A", "B", "C"),
  rank = c(3, 4, 1, 2),
  X = matrix(rnorm(4), ncol = 1),
  H = matrix(rnorm(4), ncol = 1),
  S = diag(runif(4))
)

bbt(
  rank = rank_i,
  sig_ij = sig_ij / 60,
  mi_ij = mi_ij / 60
)

gonzo(
  rank = c(3, 4, 1, 2),
  mi_i = c(1500, 1400, 1550, 1700),
  sig_i = c(200, 30, 100, 300)
)

glicko(
  teams = c("A", "B", "C", "D"),
  rank = c(3, 4, 1, 2),
  days = c(0, 0, 0, 0),
  r = c(1500, 1400, 1550, 1700),
  rd = c(200, 30, 100, 300),
  init_r = 1500,
  init_rd = 100
)

glicko2(
  teams = c("A", "B", "C", "D"),
  rank = c(3, 4, 1, 2),
  days = c(0, 0, 0, 0),
  r = c(1500, 1400, 1550, 1700),
  rd = c(200, 30, 100, 300),
  sig = c(.06, .06, .05, .07),
  tau = .5,
  init_r = 1500,
  init_rd = 100
)