Edit on GitHub

The Player class

The features of the agents are defined within class Player: the prior Gaussian distribution characterized by the mean (mu) and the standard deviation (sigma), the standard deviation of the performance (beta), and the dynamic uncertainty of the skill (gamma).

TrueSkillThroughTime.Player — Type

The Player class is used to define the features of the agents. We can create objects by indicating the parameters in order or by mentioning their names. 

Player(prior::Gaussian=Gaussian(MU,SIGMA), beta::Float64=BETA, gamma::Float64=GAMMA)
Player(;prior::Gaussian=Gaussian(MU,SIGMA), beta::Float64=BETA, gamma::Float64=GAMMA)
  • prior is the prior belief distribution of skill hypotheses
  • beta is the standar deviation of the agent performance
  • gamma is the uncertainty (standar deviation) added to the estimates as time progresses

The default value of MU, SIGMA, BETA and GAMMA are 

julia> a1 = ttt.Player()Player(Gaussian(mu=0.0, sigma=6.0), beta=1.0, gamma=0.03)
julia> a2 = ttt.Player(ttt.Gaussian(0.0, 1.0))Player(Gaussian(mu=0.0, sigma=1.0), beta=1.0, gamma=0.03)

We can also create special players who have non-random performances (beta=0.0), and whose skills do not change over time (gamma=0.0).

julia> a3 = ttt.Player(beta=0.0, gamma=0.0)Player(Gaussian(mu=0.0, sigma=6.0), beta=0.0, gamma=0.0)
julia> a3.beta0.0
julia> a3.gamma0.0

Performance

The performances $p$ are random variables around their unknown true skill $s$,

$p \sim \mathcal{N}(s,\beta^2)$

julia> ttt.performance(a2)Gaussian(mu=0.0, sigma=1.414214)
julia> ttt.performance(a3)Gaussian(mu=0.0, sigma=6.0)