makeWAR: Compute the runs above average (RAA) for each play.
In beanumber/openWAR: Machinery for Analyzing Baseball Data and Computing WAR

Description Usage Arguments Details Value Examples

Computes runs above average (RAA) for each player involved in each play of the GameDayPlays object.

makeWAR(x, models = list(), verbose = TRUE, low.memory = TRUE,
  step = FALSE, ...)

## S3 method for class 'GameDayPlays'
makeWAR(x, models = list(), verbose = TRUE,
  low.memory = TRUE, step = FALSE, ...)

## S3 method for class 'GameDayPlays_Runs'
makeWAR(x, models = list(), verbose = TRUE,
  low.memory = TRUE, step = FALSE, ...)

## S3 method for class 'GameDayPlays_Fielding'
makeWAR(x, models = list(), verbose = TRUE,
  low.memory = TRUE, step = FALSE, ...)

## S3 method for class 'GameDayPlays_Pitching'
makeWAR(x, models = list(), verbose = TRUE,
  low.memory = TRUE, step = FALSE, ...)

## S3 method for class 'GameDayPlays_Offense'
makeWAR(x, models = list(), verbose = TRUE,
  low.memory = TRUE, step = FALSE, ...)

## S3 method for class 'GameDayPlays_Baserunning'
makeWAR(x, models = list(),
  verbose = TRUE, low.memory = TRUE, step = FALSE, ...)

## S3 method for class 'GameDayPlays_Batters'
makeWAR(x, models = list(), verbose = TRUE,
  low.memory = TRUE, step = FALSE, ...)

`x`	An object of class `GameDayPlays`_, where can be the output of and of the seven steps listed below.
`models`	A named list of models, each with a `predict` method. See Details.
`verbose`	A `logical` indicating whether you want various messages and information to be displayed during the computation
`low.memory`	A `logical` indicating whether to conserve memory by not storing the model objects.
`step`	A `logical` indicating whether the function should stop after one of the seven steps.
`...`	currently ignored

Within a GameDayPlays object, each row consists of a single plate appearance and contains information about the batter, all of the baserunners, the pitcher, and all of the fielders on the field during the plate appearance. The total value of the play as determined by the change in the run expectancy matrix from the beginning of the plate appearence to the end of the plate appearance is partitioned across all players involved in the play on offense, and that same value (with the opposite sign) is partitioned across the pitcher and all of the fielders. Thus for every single plate appearance a runs above average (RAA) value is assigned to every player involved in the play. If no models argument is supplied, then all models necessary for the computation of openWAR will be generated on the data set given. The output of this function is then used in the function getWAR to calculate a Wins Above Replacement (WAR) value for each player.

If verbose == TRUE, then various pieces of information will be displayed during the computation.

The makeWAR algorithm consists of seven steps, each of which appends columns to the original data frame.

Compute the change in run expectancy (δ) on each plate appearance
Partition -δ among the fielders using makeWARFielding
Partition the remaining δ for the pitcher using getModelPitching
Adjust δ for cicumstance, and attribute it to the offense using getModelOffense
Partition some of δ for the baserunners, using getModelBaserunning
Attribute the rest of δ to the batter, and adjust using getModelBatting
Parition the baserunning δ to the various baserunners, using makeWARBaserunning

Elements of models:

run-expectancy: a model for assigning a run expectancy value to any of the 24 (base,out) states. Variables must be startCode [0-7] and startOuts [0-2]
pitching: a model for the expected outcome of a plate appearance attributable to the pitcher. Variables must be venueId, throws [L/R], and stands [L/R]
offense: a model for the expected outcome of a plate appearance attributable to the offense. Variables must be venueId, throws [L/R], and stands [L/R]
baserunning: a model for the expected contribution of the baserunners to a plate appearance. Variables must be event (the type of batting event), startCode [0-7], and startOuts [0-2]
batting: a model for the expected contribution of the batter to a plate appearance. Variables must be batterPos (the defensive position of the batter)

An object of class openWARPlays which is a list containing the following:

dataA data.frame containing the original data along with appended rows containing the RAA values for each player involved in a plate appearance. NULL if low.memory is TRUE.
models.usedA list containing all of the model information for each of the models used in computing RAA. NULL if low.memory is TRUE.
openWARPlaysA data.frame of class openWARPlays containing only the columns necessry for input into the getWAR function.

## Not run: 
res <- makeWAR(May)
summary(getWAR(res))

print(object.size(res), units = "Mb")
res <- makeWAR(May, low.memory = FALSE)
print(object.size(res), units = "Mb"
names(res$models.used)

## End(Not run)
res <- makeWAR(May, step = TRUE)
print(object.size(res), units = "Mb")
## Not run: 
res <- makeWAR(May, step = TRUE)
print(object.size(res), units = "Mb")

res2 <- makeWAR(res, step = TRUE)
print(object.size(res2), units = "Mb")

res3 <- makeWAR(res2, step = TRUE)
print(object.size(res3), units = "Mb")

res4 <- makeWAR(res3, step = TRUE)
print(object.size(res4), units = "Mb")

res5 <- makeWAR(res4, step = TRUE)
print(object.size(res5), units = "Mb")

res6 <- makeWAR(res5, step = TRUE)
print(object.size(res6), units = "Mb")

res7 <- makeWAR(res6, step = TRUE)
print(object.size(res7), units = "Mb")

## End(Not run)