Add whiff, csw, swing

convert.py

@@ -95,7 +95,7 @@ presult = {
     122: 'Catcher interference',
     123: 'Uncaught third strike',
     124: 'Sacrifice hit error',
-    125: 'Sacrifice fly, error',
+    125: 'Sacrifice fly error',
     126: "Fielder's choice",
     128: "Sacrifice fielder's choice",
     129: 'Bunt strikeout',

data.py

@@ -1,5 +1,4 @@
 import polars as pl
-from glob import glob
 import os
 from tqdm.auto import tqdm
 
@@ -8,6 +7,7 @@ from convert import aux_global_id_to_code, presult, ball_kind, ball_kind_code, l
 DATA_PATH = os.path.expanduser('~/Documents/npb_data_collector/npb')
 # SEASONS = list(range(2021, 2025+1))
 SEASONS = [2021, 2022, 2023, 2024, 2025]
+# SEASONS = [2024]
 
 data_df = pl.DataFrame()
 text_df = pl.DataFrame()
@@ -86,8 +86,6 @@ aux_df = (
     )
 )
 
-data_df = data_df
-
 data_df = (
     data_df
     .with_columns(
@@ -153,16 +151,28 @@ data_df = (
 
         pl.col('x').add(-100).mul(-1),
         pl.col('y').neg().add(250),
-        pl.col('presult').replace_strict(presult),
+        pl.col('presult').alias('presult_id'),
         pl.col('ballKind').replace_strict(ball_kind),
         pl.col('ballKind').replace_strict(ball_kind_code).alias('ballKind_code'),
         pl.col('batLR').replace_strict(lr),
+        pl.col('date').str.to_date('%Y%m%d'),
         
         pl.when(pl.col('GameKindName').str.contains('Regular Season') | (pl.col('GameKindName') == 'Interleague'))
         .then(pl.lit('Regular Season'))
         .when(~pl.col('GameKindName').is_in(['Spring Training', 'All-Star Game']))
         .then(pl.lit('Postseason'))
         .otherwise('GameKindName')
-        .alias('coarse_game_kind')
+        .alias('coarse_game_kind'),
+    )
+    .with_columns(
+        pl.col('presult_id').replace_strict(presult).alias('presult')
+    )
+    .with_columns(
+        pl.col('presult').is_in(['None', 'Balk', 'Batter interference', 'Catcher interference', 'Pitcher delay', 'Intentional walk', 'Unknown']).not_().alias('pitch'),
+        pl.col('presult').is_in(['Swinging strike', 'Swinging strikeout']).alias('whiff'),
+    )
+    .with_columns(
+        (pl.col('pitch') & pl.col('presult').is_in(['Hit by pitch', 'Sacrifice bunt', 'Sacrifice fly', 'Looking strike', 'Ball', 'Walk', 'Looking strikeout', 'Sacrifice hit error', 'Sacrifice fly error', "Sacrifice fielder's choice", 'Bunt strikeout']).not_()).alias('swing'),
+        (pl.col('whiff') | pl.col('presult').is_in(['Looking strike', 'Uncaught third strike', 'Looking strikeout'])).alias('csw')
     )
 )