Skip to contents

Infer the role of each player

Source: R/rotations.R
ov_infer_player_roles.Rd

Infer the role of each player

Usage

ov_infer_player_roles(
  x,
  target_team,
  method,
  fall_back = TRUE,
  setter_tip_codes = c("PP", "PR", "P2")
)

Arguments

x

: a datavolley object (as returned by datavolley::dv_read()), a list of datavolley objects, or the plays component of a datavolley object

target_team

string or function: team to report on. If this is a function, it should return TRUE when passed the target team name

method

string: one of

  • "meta" - rely on player metadata, i.e. the team rosters

  • "SHM" - assume a setter-hitter-middle rotation order. When the setter is in position 1 the outside hitters are in 2 and 5, and middles in 3 and 6

  • "SMH" - setter-middle-hitter rotation order. When the setter is in position 1 the middles are in 2 and 5, and outside hitters in 3 and 6

  • or "data" - try and identify player roles based on the ball touches that each player makes. Note that this assumes certain play characteristics: outsides hit pipes and in front row hit from the left except in P1 reception, opposites hit from the right (left in P1 reception), middles hit quick balls or slides or from zone 3 Method "meta" is the default if a datavolley object or list of objects is provided. Note that "SHM" and "SMH" cannot identify liberos, because they do not appear in the point-by-point on-court lineups.

fall_back

logical: if TRUE and method is "meta" and x is a single datavolley object BUT player roles are not provided in the DataVolley file metadata section, fall back to method "data"

setter_tip_codes

character: vector of attack combination codes that correspond to setter tips and other attacks that do not tend to be made from a consistent start zone. Only relevant if method is "data"

Value

A data.frame with columns player_id, role, and score (the approximate confidence in the role assignment, in the range 0 to 1)

Details

The inference procedure assumes that each player plays the same role across the entire data set x. If x is drawn from multiple matches, and a player changes role from match to match, it will likely give misleading results for that player. In that case, it is probably better to run this inference on each match separately.

Examples

x <- ovdata_example("mlafin_braslovce_nkbm", as = "parsed")
## guess roles according to the actions that the players made
rx <- ov_infer_player_roles(x, target_team = "Nova KBM Branik", method = "data")