write_bool = any(x.lower() == "write" for x in sys.argv)
new_stats_bool = any(x.lower() == "new" for x in sys.argv)
# Examines the "new" argument. If it exists, use the new scraped stats and combine them. If it does not exist,use the data from previously, the R-code-based one that combines the stats for each year.
if new_stats_bool:
# This must be run first to give the pickle files access to what they need.
import nfl_convertPlayerGameStats
# Make a playerstats file. Read in all seasons and then concatenate them vertically.
# FIXME: Keeping this set to 2018 for now
seasons = range(1999, 2018 + 1)
outlist = []
for season in seasons:
playerstatfile = Path(
"data_raw", "nfl", f"nfl-player-game-statistics{season}.csv")
playerstats_season = readplayergamestats(playerstatfile)
outlist.append(playerstats_season)
playerstats = pd.concat(outlist)
else:
playerstatfile = Path("data_raw", "nfl", cfg["playerstatfile"])
playerstats = readplayergamestats(playerstatfile)
if any(x.lower() == "qb" for x in sys.argv) or all_bool:
# Subset of dataframe
meet_cond = playerstats[(playerstats["Pos"].isin(["QB"]))].copy()
# Calculated variables
# meet_cond["calc_YFS"] = meet_cond["Rush Yard"] + meet_cond["Rec Yards"]
meet_cond["calc_pass_comp"] = (meet_cond["Pass Comp"] / meet_cond["Pass Att"]) * 100
meet_cond["calc_yard_per_att"] = meet_cond["Pass Yard"] / meet_cond["Pass Att"]
meet_cond["calc_yard_per_comp"] = meet_cond["Pass Yard"] / meet_cond["Pass Comp"]
from tqdm import tqdm
tqdm.write("Beginning ComputeElo")
seasons = tqdm(seasons)
for season in seasons:
seasons.set_description(f"s: {season}")
# Read in files for the current season
playerstatfile = Path("data_raw", "cfb",
f"{playerstatroot}{season}{extension}")
teamstatfile = Path("data_raw", "cfb",
f"{teamstatroot}{season}{extension}")
teamstats = eu.readteamgamedata(teamstatfile)
# Read in and merge player stats with demographics/position
playerstats = eu.readplayergamestats(playerstatfile)
dates = [int(x) for x in sorted(list(playerstats["gamedate"].unique()))]
# Filter based on stats
# FUTURE: change here to process based on player's position.
# FUTURE: [player["position"].rsplit("/")] -- returns a list of positions the player has.
if processRBs:
# RBs - Filter to only players with at least 1 Rush Attempt
rbstats = playerstats[playerstats["Rush Att"] > 0].copy()
if processWRs:
# WRs/TEs - Filter to only players with at least 1 Catch
wrstats = playerstats[playerstats["Rec"] > 0].copy()
if processQBs:
from eloUtilities import readplayergamestats
# Get filename from config file
with open(Path("elo_config.json")) as file:
cfg = json.load(file)["findStats"]["college"]
extension = ".csv"
# Seasons to run
seasons = range(2005, 2019 + 1)
# Read in files
all_seasons = [
readplayergamestats(
Path("..", "data_raw", "cfb",
f"{cfg['playerstatfile']}{season}{extension}"))
for season in seasons
]
# Combine all the dataframes into 1 to calculate and find medians for
playerstats = pd.concat(all_seasons, axis=0, ignore_index=True)
# Subset of dataframe
meet_cond = playerstats[playerstats["Rush Att"] > 10].copy()
# Calculated variables
meet_cond["calc_YPC"] = meet_cond["Rush Yard"] / meet_cond["Rush Att"]
# Return median for selected variables
# print("recs:", meet_cond["Rec"].median())