Function to get time-sequenced spatial locations of a baseball (optionally summarized over an arbitrary set of variables) using PITCHf/x

getLocations.R

if (!require("dplyr")) install.packages("dplyr")
if (!require("tidyr")) install.packages("tidyr")
if (!require("pitchRx")) install.packages("pitchRx")

getLocations <- function(dat, ..., summarise = TRUE) {
  # select and group by columns specified in ...
  tb <- dat %>%
    select(..., x0:az) %>%
    group_by(...)
  vars <- as.character(attr(tb, "vars"))
  if (summarise) {
    # average the PITCHf/x parameters over variables specified in ...
    labs <- attr(tb, "labels")
    tb <- tb %>% summarise_each(funs(mean))
  } else {
    # another (more complex way to get variables names)
    # vars <- as.character(as.list(match.call(expand.dots = FALSE))$...)
    dat$pitch_id <- seq_len(nrow(dat))
    vars <- c(vars, "pitch_id")
    labs <- dat[vars]
  }
  # returns 3D array of locations of pitches over time
  value <- pitchRx::getSnapshots(as.data.frame(tb))
  idx <- labs %>% unite_("id", vars, sep = "@&")
  dimnames(value) <- list(idx = as.data.frame(idx)[, 1],
                          frame = seq_len(dim(value)[2]),
                          coordinate = c("x", "y", "z"))
  # tidy things up in a format that ggplot would expect
  value %>% as.tbl_cube() %>% as.data.frame() %>% rename_(value = ".") %>% 
    mutate(idx = as.character(idx)) %>% 
    separate(idx, vars, sep = "@&") %>%
    spread(coordinate, value)
}

I think there is a bug with something related to the group_by(...). When I run
dat <- getLocations(pitches, pitcher_name, pitch_type, summarise = TRUE)
I am getting an error saying "Error: index out of bounds"

Thanks! This should work now.

Hi Carson,
I downloaded Pitchfx data for all 2014 and joined atbat and pitch tables for Yu Darvish. But when I run getLocations function, R keeps showing Error in 0:(nplots - 1) : NA/NaN argument. Do you have any idea why this function doesn't work?
Thanks so much!