100 m pace equivalents

100m-pace.R

# Context: https://bsky.app/profile/kjhealy.bsky.social/post/3kywqjin76i2r

library(rvest)
library(tinyplot)
tpar(grid = TRUE, las = 1, pch = 16)

# read the wiki page
wp = read_html("https://en.wikipedia.org/wiki/List_of_world_records_in_athletics") 

# select the table of interest
dat = wp |>
  html_element("div+ .wikitable :nth-child(1)") |> ## select table element
  html_table()                                     ## convert to data frame

# subset rows of interest
dat = dat |>
  subset(grepl(" m$| km$|Mile$|marathon$|^Marathon", Event)) |>
  subset(!grepl("relay", Event)) |>
  subset(!(Event %in% c("1000 m", "2000 m", "Mile"))) |> # non-standard events
  subset(Pts!="N/A")

# futzing to convert string times into consistent second measurements
dat = dat |> 
  transform(sec = ifelse(!grepl(":", Perf.), paste0("0:", Perf.), Perf.)) |>
  transform(sec = ifelse(grepl("\\.", sec), paste0("0:", sec), sec)) |>
  transform(sec = ifelse(!grepl("\\.", sec), paste0(sec, ".00"), sec)) |>
  transform(sec = as.numeric(as.difftime(sec, format = "%H:%M:%OS")))

# convert events into consistent m(etre) increments
dat = dat |>
  transform(m = as.numeric(gsub(",| m$", "", Event))) |>
  transform(m = ifelse(
    Event=="Mile",
    1600,
    ifelse(
      grepl("Half marathon", Event),
      42195/2,
      ifelse(
        grepl("Marathon", Event),
        42195,
        m
      )
    )
  ))

# 100m pace
dat = dat |>
  transform(pace_100m = sec/m*100)

# quick plot
plt(pace_100m ~ m, dat, type = "b")

# looks like a linear-log model
mod = lm(pace_100m ~ log(m), dat)
pred = predict(mod, interval = "confidence")
dat = cbind(dat, pred)

# add predictions
with(
  dat,
  plt(
    x = m, y = fit, ymin = lwr, ymax = upr,
    type = "ribbon", col = "hotpink",
    log = "x",
    xlab = "Event distance (m), log scale",
    ylab = "Average 100m pace (sec)",
    main = "Who is the 'fastest'?"
  )
)
with(dat, plt(m, pace_100m, type = "b", add = TRUE))