Import Libraries.¶

In [ ]:
library(tidyverse)
library(baseballr)
library(ggthemes)
library(ggimage)
library(gt)
library(readr)
options(scipen = 9999)

Pull Statcast Data.¶

In [ ]:
start_date <- as.Date('2022-04-07', '%Y-%m-%d')
end_date <- as.Date('2022-10-05', '%Y-%m-%d')

date_seq <- seq(from=start_date, to=end_date, by="day")
date_seq <- as.character(date_seq)
In [ ]:
# 30 min runtime for full season.
statcast_df <- data.frame()
for (date in date_seq){
    suppressWarnings({
        statcast <- statcast_search_pitchers(start_date=date, end_date=date)
    })
    if (nrow(statcast) > 0){
        statcast_df <- rbind(statcast_df, statcast)
    }
}

Pull MLB Stats API Data.¶

In [ ]:
schedule <- mlb_schedule(season="2022") %>%
    filter(status_detailed_state != "Postponed") %>%
    filter(status_detailed_state != "Cancelled") %>%
    filter(series_description == "Regular Season") %>%
    group_by(game_pk) %>%
    summarize(count=n())

games_list <- as.list(schedule$game_pk)
In [ ]:
# 120 min runtime for full season.
pbp_df <- data.frame()
for (i in games_list){
    pbp <- mlb_pbp(i) %>%
        filter(isPitch == "TRUE")
    pbp_df <- bind_rows(pbp_df, pbp)
}

Merge Datasets.¶

In [ ]:
statcast_df$at_bat_number <- sprintf("%02d", statcast_df$at_bat_number)
statcast_df$pitch_number <- sprintf("%02d", statcast_df$pitch_number)
statcast_df$pitch_id <- paste(statcast_df$game_pk, statcast_df$at_bat_number, statcast_df$pitch_number, sep = "-")

pbp_df$atBatIndex <- sprintf("%02d", strtoi(pbp_df$atBatIndex) + 1)
pbp_df$pitchNumber <- sprintf("%02d", strtoi(pbp_df$pitchNumber))
pbp_df$pitch_id <- paste(pbp_df$game_pk, pbp_df$atBatIndex, pbp_df$pitchNumber, sep = "-")
In [ ]:
joined_df <- merge(pbp_df, statcast_df, by="pitch_id", all=TRUE)
joined_df <- subset(joined_df, select=-reviewDetails.additionalReviews)

Save Merged Dataset to CSV.¶

In [ ]:
# 3 min runtime for full season. 1.2GB file size.
write.csv(joined_df,"C:\\Users\\chris\\Documents\\Datasets\\BaseballR\\2022\\2022merged.csv", row.names = FALSE)