npb_data_app / pitch_leaderboard.py
patrickramos's picture
Start pitch leaderboard
8353665
raw
history blame
3.49 kB
import gradio as gr
import polars as pl
from datetime import datetime
from itertools import chain
from data import data_df
from stats import compute_pitch_stats, filter_data_by_date_and_game_kind
from convert import ball_kind
STATS = ['Count', 'Usage', 'SwStr%', 'Whiff%', 'CSW%', 'GB%', 'FB%', 'LD%']
PCT_STATS = ['Usage', 'SwStr%', 'Whiff%', 'CSW%', 'GB%', 'FB%', 'LD%']
STATS_WITH_PCTLS = ['SwStr%', 'Whiff%', 'CSW%', 'GB%', 'FB%', 'LD%']
def gr_create_pitch_leaderboard(start_date, end_date, min_pitches, include_pitches):
data = data_df.filter(pl.col('ballKind_code') != '-')
data = filter_data_by_date_and_game_kind(data, start_date=start_date, end_date=end_date, game_kind='Regular Season')
both, left, right = [
(
compute_pitch_stats(df, player_type='pitcher', min_pitches=min_pitches, pitch_class_type='specific')
.filter(pl.col('qualified') & (pl.col('ballKind').is_in(include_pitches)))
.drop('qualified')
.rename({'pitcher_name': 'Pitcher', 'count': 'Count', 'usage': 'Usage', 'ballKind': 'Pitch', 'general_ballKind': 'Pitch (General)'} | {f'{stat}_pctl': f'{stat} (Pctl)' for stat in STATS_WITH_PCTLS})
.with_columns(
pl.col(stat).mul(100).round(1)
for stat in PCT_STATS + [f'{stat} (Pctl)' for stat in STATS_WITH_PCTLS]
)
[['pitId', 'ballKind_code', 'Pitcher', 'Pitch', 'Pitch (General)', 'Count', 'Usage'] + STATS_WITH_PCTLS]
)
for df
in [data, data.filter(pl.col('batLR') == 'l'), data.filter(pl.col('batLR') == 'r')]
]
pitch_stats = (
both
.join(left, on=['pitId', 'ballKind_code'], suffix=' (LHH)', how='full')
.join(right, on=['pitId', 'ballKind_code'], suffix=' (RHH)', how='full')
.drop('pitId', 'ballKind_code', *list(chain.from_iterable([[f'{col} ({handedness}HH)' for col in ['pitId', 'ballKind_code', 'Pitcher', 'Pitch', 'Pitch (General)']] for handedness in ('L', 'R')])))
)
return pitch_stats
def create_pitch_leaderboard():
now = datetime.now()
start_datetime_init = datetime(now.year, 1, 1)
end_datetime_init = now
pitch_types = [pitch_type for pitch_type in ball_kind.values() if pitch_type != '-']
with gr.Blocks() as app:
gr.Markdown('# Pitch Leaderboard')
with gr.Row():
start_date = gr.DateTime(start_datetime_init, include_time=False, type='datetime', label='Start')
end_date = gr.DateTime(end_datetime_init, include_time=False, type='datetime', label='End')
min_pitches = gr.Number(100, label='Min. Pitches', precision=0, minimum=0)
with gr.Row():
include_pitches = gr.CheckboxGroup(pitch_types, value=pitch_types, label='Pitches', scale=2)
# with gr.Column(scale=1):
# all_pitches = gr.Checkbox(label='Select all pitches')
search = gr.Button('Search')
leaderboard = gr.DataFrame(
# gr_create_pitch_leaderboard(start_date=start_date_init, end_date=end_date_init, min_pitches=100),
column_widths=[200]*3 + [100]*(3*len(STATS)),
show_copy_button=True,
show_search=True,
pinned_columns=2
)
search.click(gr_create_pitch_leaderboard, inputs=[start_date, end_date, min_pitches, include_pitches], outputs=leaderboard)
# include_pitches.input(lambda _pitch_types: (_pitch_types == pitch_types), inputs=include_pitches, outputs=all_pitches)
# all_pitches.input(lambda _all_pitches : pitch_types, inputs=all_outputs=all_pitches)
return app
if __name__ == '__main__':
app = create_pitch_leaderboard()
app.launch()