Update missing linebacker ids from 2022-2023

[jak3sch]

I found a bug in the data from yesterday.

load_ff_playerids() only has only LB as position tag but load_rosters_weekly() differentiates in LB, ILB, MLB and OLB. That's why some players couldn't be joined.

I ran the same code as yesterday with one addition (look for '#NEW LINE') to use LB as position tag for all LBs.

library(nflreadr)
library(tidyverse)
library(dplyr)

#Get all rookies from 2022 and 2023 with MFL_ids
ff_playerids <- nflreadr::load_ff_playerids() %>%
  dplyr::filter(draft_year >= 2022)

#Get all Rookies from 2022 and their gsis_ids
rosters_weekly_2022 <- nflreadr::load_rosters_weekly(seasons = 2022) %>%
  dplyr::filter(rookie_year >= 2022) %>%
  dplyr::distinct(gsis_id, .keep_all = TRUE)

#Get all rookies from 2023 and their gsis_ids
rosters_weekly_2023 <- nflreadr::load_rosters_weekly(seasons = 2023) %>%
  dplyr::filter(rookie_year >= 2022) %>%
  dplyr::distinct(gsis_id, .keep_all = TRUE)

#Get pfr_ids out of pfr dataset
pfr_ids <- nflreadr::load_snap_counts(seasons = 2022:2023) %>%
  dplyr::mutate(player = nflreadr::clean_player_names(player)) %>%
  dplyr::group_by(pfr_player_id, position) %>%
  dplyr::arrange(season, week) %>%
  dplyr::filter(dplyr::row_number() == 1) %>%
  dplyr::filter(
    # filter some ids manually
    !pfr_player_id %in% c(
      "BrowMi0a", # executive
      "JohnBr20", # LB drafted in 2006
      "AlexMa00" # DB drafted in 2014
    )
  ) %>%
  dplyr::mutate(sync_helper = paste0(season, position, gsub(" ", "", player))) %>%
  dplyr::ungroup() %>%
  dplyr::select(player, pfr_player_id, position, sync_helper) %>%
  dplyr::distinct()

gsis_to_sync <- nflreadr::load_rosters_weekly(2022:2023) %>%
  #  dplyr::filter(is.na(draft_number)) %>%
  dplyr::mutate(
    player = nflreadr::clean_player_names(full_name),
    depth_chart_position = ifelse(position == "LB", position, depth_chart_position) # NEW LINE
  ) %>%
  dplyr::group_by(gsis_id, depth_chart_position) %>%
  dplyr::arrange(season, week) %>%
  dplyr::filter(dplyr::row_number() == 1) %>%
  dplyr::mutate(sync_helper = paste0(season, depth_chart_position, gsub(" ", "", player))) %>%
  dplyr::ungroup() %>%
  dplyr::select(player, gsis_id, sync_helper) %>%
  dplyr::filter(!is.na(gsis_id))

combined_ids <- gsis_to_sync %>%
  dplyr::left_join(
    pfr_ids %>%
      dplyr::select(pfr_player_id, sync_helper),
    by = "sync_helper",
    relationship = "many-to-many"
  ) %>%
  dplyr::filter(!is.na(pfr_player_id))

gsis_to_pfr_ids <- combined_ids %>%
  dplyr::select(gsis_id,pfr_player_id) %>%
  dplyr::rename(pfr_id = pfr_player_id)  %>%
  dplyr::distinct()

# check for duplicates
test <- combined_ids %>%
  dplyr::group_by(sync_helper) %>%
  dplyr::summarise(count = dplyr::n())

#join to fill columns
joined <- ff_playerids %>%
  dplyr::left_join(rosters_weekly_2022, by = join_by(name==full_name, position==position)) %>%
  dplyr::left_join(rosters_weekly_2023, by = join_by(name==full_name, position==position))

#This produces 0 results, so we know we didn't get any gsis_ids that were wrong, everything either matches or fills in data
check_joined <- joined %>%
  dplyr::filter(gsis_id.x != gsis_id.y)

output_to_csv <- joined %>%
  dplyr::mutate(
    gsis_id = case_when(!is.na(gsis_id.x) ~ gsis_id.x,
                        !is.na(gsis_id.y) ~ gsis_id.y,
                        .default = gsis_id
    ),
    sleeper_id = case_when(!is.na(sleeper_id.x) ~ sleeper_id.x,
                           !is.na(sleeper_id.y) ~ sleeper_id.y,
                           .default = sleeper_id),
    pff_id = case_when(!is.na(pff_id.x) ~ pff_id.x,
                       !is.na(pff_id.y) ~ pff_id.y,
                       .default = pff_id),
    espn_id = case_when(!is.na(espn_id.x) ~ espn_id.x,
                        !is.na(espn_id.y) ~ espn_id.y,
                        .default = espn_id),
    yahoo_id = case_when(!is.na(yahoo_id.x) ~ yahoo_id.x,
                         !is.na(yahoo_id.y) ~ yahoo_id.y,
                         .default = yahoo_id),
    ras_id = NA,
    otc_id = NA
  ) %>%
  dplyr::filter(!is.na(sleeper_id) | !is.na(pfr_id) | !is.na(gsis_id) | !is.na(pff_id) | !is.na(yahoo_id) | !is.na(ras_id) | !is.na(otc_id)) %>%
  dplyr::select(mfl_id,sleeper_id,gsis_id,pff_id,espn_id,yahoo_id,ras_id,otc_id) %>%
  #This last part is to join in
  dplyr::left_join(gsis_to_pfr_ids, by = join_by(gsis_id)) %>%
  dplyr::arrange(mfl_id) %>%
  dplyr::mutate(count_col = paste0((838+row_number()))) %>%
  #To reorder the columns
  dplyr::select(count_col,mfl_id,sleeper_id,pfr_id,gsis_id,pff_id,espn_id,yahoo_id,ras_id,otc_id)

write.csv(output_to_csv, "new_missing_ids.csv", row.names=FALSE)

[tanho63]

Hmm. Can you combine the new_missing_ids csv with the missing_ids.csv?

[jak3sch]

I added the IDs to the existing file. One flaw I noticed: players, who didn't played in their rookie season will not be joined properly. One example is Justyn Ross. I added him manually.

I will try to check for missing players from time to time

[tanho63]

LGTM, thanks!