findK: Generate the appropriate K factor given a players score

In nmmarquez/linHierarchy: Generate linear rankings from dyadic interactions within a group

Description

Given a players score, ranges of scores to associate with a K factor and the K values to use for the given ranges the function will find the appropriate K score to use for an elo rtaing update based on Neumann et al 2011.

Usage

findK(playerScore, kScaleBounds, kFactors)

Arguments

playerScore	the score of the player in question
kScaleBounds	the player score bins to associate with K factors
kFactors	the possible K factors to be used

Details

Given a parameterization of kScaleBounds = c(-Inf, 2100, 2400, Inf) and kFactors = c(32,24,16):

- If an individuals score is less than 2100 the kFactor assigned is 32

- Between 2100 and 2400 the kFactor assigned is 24

- A score above 2400 the the kFactor assigned is 16

Value

the players associated K factor

Note

In order to ensure a proper function call the kScaleBounds parameter must be sorted from least to greatest and must be of a length 1 greater than the length of the kFactors parameter. In addition assigning the first element of kScaleBounds to -Inf and the last to Inf will allow the players score to always remain within an acceptable range.

References

Neumann et al (2011) Assessing Dominance Hierarchies. Animal Behaviour

Examples

# generic function calls
findK (1000, kScaleBounds = c(-Inf, Inf), kFactors = 100)
findK (1000, kScaleBounds = c(-Inf, 1100, Inf), kFactors = c(32, 16))
# may cause errors in elo algorithim if range of kscale bounds isnt all real
# numbers but you can still call it this way
findK (1000, kScaleBounds = c(0, 500, 100, Inf), kFactors = c(32, 16, 8))

nmmarquez/linHierarchy documentation built on May 23, 2019, 9:28 p.m.