Update app.py
Browse files
app.py
CHANGED
|
@@ -92,7 +92,10 @@ from datasets import load_dataset
|
|
| 92 |
season = 2025
|
| 93 |
level = 'mlb'
|
| 94 |
# dataset = load_dataset('TJStatsApps/mlb_data', data_files=[f'{level}_pitch_data_{season}.csv' ])
|
| 95 |
-
|
|
|
|
|
|
|
|
|
|
| 96 |
# dataset_train = dataset['train']
|
| 97 |
df = dataset.drop_duplicates(subset=['play_id'],keep='last')
|
| 98 |
df['batter_name_team'] = df['batter_name'] + ' - ' + df['batter_team']
|
|
|
|
| 92 |
season = 2025
|
| 93 |
level = 'mlb'
|
| 94 |
# dataset = load_dataset('TJStatsApps/mlb_data', data_files=[f'{level}_pitch_data_{season}.csv' ])
|
| 95 |
+
from datasets import load_dataset
|
| 96 |
+
# load the dataset file from your repo
|
| 97 |
+
ds = load_dataset("TJStatsApps/mlb_data", data_files=f"data/{level}_pitch_data_{season}.parquet")
|
| 98 |
+
df = ds["train"].to_pandas() # or to_polars() if you want Polars
|
| 99 |
# dataset_train = dataset['train']
|
| 100 |
df = dataset.drop_duplicates(subset=['play_id'],keep='last')
|
| 101 |
df['batter_name_team'] = df['batter_name'] + ' - ' + df['batter_team']
|