scrape statcast expected stats

scrape_statcast_expected_stats.R

library(dplyr)
library(ggplot2)
library(rvest)
library(jsonlite)

scrape_statcast_expected_stats = function(year=2018, min_pa=25) {
  url = sprintf("https://baseballsavant.mlb.com/expected_statistics?type=batter&year=%s&position=&team=&min=%d", year, min_pa)
  h = xml2::read_html(url) 
  s = html_nodes(h, "script")[[10]]
  sm = str_match(html_text(s), "data\\s+=\\s+(\\[.+\\])")
  sdf = jsonlite::fromJSON(sm[[2]])
  
  numeric_cols = 
    c("ba", "est_ba", 
      "slg", "est_slg", "woba", "est_woba", "wobacon", "est_wobacon", 
      "wobacon_minus_est_wobacon_diff", "slg_minus_est_slg_diff", "woba_minus_est_woba_diff", 
      "ba_minus_est_ba_diff", "est_wobacon_minus_wobacon_diff", "est_ba_minus_ba_diff", 
      "est_woba_minus_woba_diff", "est_slg_minus_slg_diff")
  
  integer_cols = 
    c("year", "team_id", "pa", "bip", "age", "position_name", "pos", "player_id")
  
  for (d in numeric_cols) {
   sdf[[d]] = as.numeric(as.character(sdf[[d]])) 
  }
  
  for (d in integer_cols) {
    sdf[[d]] = as.integer(as.character(sdf[[d]])) 
  }
  
  sdf
}

x_df = scrape_expected(min_pa=1)
x_df %>% ggplot(aes(x= pa, y=woba - est_woba)) + geom_point() + xlim(1, 200)  + ylim(-0.2, 0.15) + geom_smooth() + theme_minimal(base_size = 16) + geom_hline(yintercept=0) + labs(x="PA", y="wOBA - est_woba", title="Expected vs Observed wOBA - 2018")