patrickramos commited on
Commit
0c44916
·
1 Parent(s): 75d7643

Allow dataframe caching in data.py

Browse files
Files changed (3) hide show
  1. data/__init__.py +11 -0
  2. data.py → data/data.py +4 -2
  3. data/seasons.py +1 -0
data/__init__.py ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import os
2
+ import polars as pl
3
+
4
+ from .seasons import SEASONS
5
+
6
+ assert (USE_CACHE := os.getenv('USE_CACHE', 'False')) in ('True', 'False')
7
+
8
+ if USE_CACHE == 'True':
9
+ data_df = pl.read_parquet('data.parquet')
10
+ else:
11
+ from .data import data_df
data.py → data/data.py RENAMED
@@ -20,6 +20,7 @@ from convert import (
20
  lr,
21
  game_kind
22
  )
 
23
 
24
  DATA_PATH = './files'
25
 
@@ -41,8 +42,6 @@ else:
41
 
42
  print(DATA_PATH)
43
 
44
- SEASONS = [2021, 2022, 2023, 2024, 2025]
45
-
46
  data_df = pl.DataFrame()
47
  text_df = pl.DataFrame()
48
  aux_df = pl.DataFrame()
@@ -471,5 +470,8 @@ data_df = (
471
  how='left'
472
  )
473
  )
 
 
 
474
  if __name__ == '__main__':
475
  breakpoint()
 
20
  lr,
21
  game_kind
22
  )
23
+ from .seasons import SEASONS
24
 
25
  DATA_PATH = './files'
26
 
 
42
 
43
  print(DATA_PATH)
44
 
 
 
45
  data_df = pl.DataFrame()
46
  text_df = pl.DataFrame()
47
  aux_df = pl.DataFrame()
 
470
  how='left'
471
  )
472
  )
473
+
474
+ data_df.write_parquet('data.parquet')
475
+
476
  if __name__ == '__main__':
477
  breakpoint()
data/seasons.py ADDED
@@ -0,0 +1 @@
 
 
1
+ SEASONS = [2021, 2022, 2023, 2024, 2025]