extract_elo {EloRating}

extract Elo ratings from elo object

Description

extract Elo ratings from elo object

Usage

extract_elo(
  eloobject,
  extractdate = eloobject$misc["maxDate"],
  standardize = FALSE,
  IDs = NULL,
  NA.interpolate = FALSE,
  daterange = 1
)

Arguments

eloobject

result from elo.seq

extractdate

character, date on which Elo ratings should be obtained, defaults to the last day in the data set

standardize

logical, should the returned ratings be scaled between 0 and 1. Default is FALSE. See scale_elo

IDs

character, specify IDs for which ratings are returned. By default, returns all that were present on the date or at least on one day of the date range

NA.interpolate

if FALSE (default), the last known rating is returned, which might not be from the specified date itself (but older). If TRUE, ratings on days without observations are linearly interpolated between days with known ratings (i.e. dates with observed interactions)

daterange

if averaged ratings are desired, supply here the number of days from
extractdate - 1. By default (daterange = 1), the ratings of the single extractdate are returned. daterange = 2 produces average ratings from extractdate and the day after, and so on...

Details

extractdate can be also a vector of dates. In this case, the IDs argument has to be either a vector of length 1 (i.e. a single individual) or a vector of the same length as extractdate. In the first case, the ratings for the same individual are returned on the dates specified in extractdate. In the second case, dates and IDs are matched, i.e. the rating of the individual on that date is returned in the same order as the dates/IDs vectors.

Value

named (IDs) vector of (average) Elo ratings, or an unnamed vector of ratings (if length of extracte is larger than 1)

Author(s)

Christof Neumann

Examples

data(adv)
SEQ <- elo.seq(winner=adv$winner, loser=adv$loser, Date=adv$Date)
extract_elo(SEQ, "2010-01-30")
extract_elo(SEQ, "2010-01-30", standardize=TRUE)

# same ratings (regardless of NA.interpolate),
# since "g" was observed on both days
extract_elo(SEQ, "2010-01-29", IDs="g")
extract_elo(SEQ, "2010-01-29", IDs="g", NA.interpolate=TRUE)

extract_elo(SEQ, "2010-01-31", IDs="g")
extract_elo(SEQ, "2010-01-31", IDs="g", NA.interpolate=TRUE)

# different ratings (depending on NA.interpolate),
# since "g" was not observed that day
extract_elo(SEQ, "2010-01-30", IDs="g")
extract_elo(SEQ, "2010-01-30", IDs="g", NA.interpolate=TRUE)

extract_elo(SEQ, "2010-01-10", daterange=5)
extract_elo(SEQ, "2010-01-10", daterange=5, NA.interpolate=TRUE)

# and for multiple dates and a single IDs
dates <- sample(adv$Date, size = 10, replace = TRUE)
ids <- "b"
extract_elo(eloobject = SEQ, extractdate = dates, standardize = FALSE, IDs = ids)

# and for multiple dates and IDs
dates <- sample(adv$Date, size = 10, replace = TRUE)
ids <- sample(colnames(advpres)[2:8], size = 10, replace = TRUE)
extract_elo(eloobject = SEQ, extractdate = dates, standardize = FALSE, IDs = ids)