cbratkovics/nba-stat-predictor
Tabular Regression • Updated
Dataset Viewer
The dataset viewer is not available for this split.
Cannot load the dataset split (in streaming mode) to extract the first rows.
Error code: StreamingRowsError
Exception: CastError
Message: Couldn't cast
date: timestamp[s]
run_date: timestamp[s]
model_id: string
started_at: timestamp[s]
limits: struct<max_tool_calls: int64, max_turns_after_tools: int64, wall_clock_seconds: double, temperature: (... 7 chars omitted)
child 0, max_tool_calls: int64
child 1, max_turns_after_tools: int64
child 2, wall_clock_seconds: double
child 3, temperature: int64
status: string
error: null
tool_calls_made: int64
latency_ms: int64
prefetch: list<item: struct<id: string, name: string, args: struct<date: timestamp[s], days: int64>, result: s (... 1566 chars omitted)
child 0, item: struct<id: string, name: string, args: struct<date: timestamp[s], days: int64>, result: struct<avail (... 1554 chars omitted)
child 0, id: string
child 1, name: string
child 2, args: struct<date: timestamp[s], days: int64>
child 0, date: timestamp[s]
child 1, days: int64
child 3, result: struct<available: bool, date: timestamp[s], reason: string, run_date: timestamp[s], stored_max_game_ (... 1465 chars omitted)
child 0, available: bool
child 1, date: timestamp[s]
child 2, reason: string
child 3, run_date: timestamp[s]
child 4, stored_max_game_date: timestamp[s]
child 5, stored_rows: int64
child 6, stored_seasons: list<item: string>
child 0, item: string
child 7, days_stale_stored: int64
child 8, dump_present: bool
child 9, dump_max_game_date: null
...
ld 1, dropped: int64
summary: string
findings: list<item: struct<kind: string, severity: string, evidence: struct<tool: string, args: struct<date: (... 343 chars omitted)
child 0, item: struct<kind: string, severity: string, evidence: struct<tool: string, args: struct<date: timestamp[s (... 331 chars omitted)
child 0, kind: string
child 1, severity: string
child 2, evidence: struct<tool: string, args: struct<date: timestamp[s], days: int64>, values: struct<available: bool, (... 267 chars omitted)
child 0, tool: string
child 1, args: struct<date: timestamp[s], days: int64>
child 0, date: timestamp[s]
child 1, days: int64
child 2, values: struct<available: bool, reason: string, days_stale: int64, mae_pts: double, mae_reb: double, mae_ast (... 190 chars omitted)
child 0, available: bool
child 1, reason: string
child 2, days_stale: int64
child 3, mae_pts: double
child 4, mae_reb: double
child 5, mae_ast: double
child 6, largest_residual_pts: double
child 7, model_pts: double
child 8, baseline_pts: double
child 9, model_reb: double
child 10, baseline_reb: double
child 11, model_ast: double
child 12, baseline_ast: double
child 13, known_missing_count: int64
child 3, text: string
generated_at: timestamp[s]
to
{'date': Value('timestamp[s]'), 'run_date': Value('timestamp[s]'), 'status': Value('string'), 'summary': Value('string'), 'findings': List({'kind': Value('string'), 'severity': Value('string'), 'evidence': {'tool': Value('string'), 'args': {'date': Value('timestamp[s]'), 'days': Value('int64')}, 'values': {'available': Value('bool'), 'reason': Value('string'), 'days_stale': Value('int64'), 'mae_pts': Value('float64'), 'mae_reb': Value('float64'), 'mae_ast': Value('float64'), 'largest_residual_pts': Value('float64'), 'model_pts': Value('float64'), 'baseline_pts': Value('float64'), 'model_reb': Value('float64'), 'baseline_reb': Value('float64'), 'model_ast': Value('float64'), 'baseline_ast': Value('float64'), 'known_missing_count': Value('int64')}}, 'text': Value('string')}), 'tool_calls_made': Value('int64'), 'model_id': Value('string'), 'latency_ms': Value('int64'), 'generated_at': Value('timestamp[s]'), 'grounding': {'checked': Value('int64'), 'dropped': Value('int64')}}
because column names don't match
Traceback: Traceback (most recent call last):
File "/src/services/worker/src/worker/utils.py", line 147, in get_rows_or_raise
return get_rows(
dataset=dataset,
...<4 lines>...
column_names=column_names,
)
File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
return func(*args, **kwargs)
File "/src/services/worker/src/worker/utils.py", line 127, in get_rows
rows_plus_one = list(itertools.islice(safe_iter(ds, dataset=dataset), rows_max_number + 1))
File "/src/services/worker/src/worker/utils.py", line 483, in safe_iter
yield from ds.decode(False) if ds.features else ds
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2951, in __iter__
for key, example in ex_iterable:
^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2461, in __iter__
for key, pa_table in self._iter_arrow():
~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2486, in _iter_arrow
for key, pa_table in self.ex_iterable._iter_arrow():
~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 547, in _iter_arrow
for key, pa_table in iterator:
^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 430, in _iter_arrow
for key, pa_table in self.generate_tables_fn(**gen_kwags):
~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 343, in _generate_tables
self._cast_table(pa_table, json_field_paths=json_field_paths),
~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 132, in _cast_table
pa_table = table_cast(pa_table, self.info.features.arrow_schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2378, in table_cast
return cast_table_to_schema(table, schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2306, in cast_table_to_schema
raise CastError(
...<3 lines>...
)
datasets.table.CastError: Couldn't cast
date: timestamp[s]
run_date: timestamp[s]
model_id: string
started_at: timestamp[s]
limits: struct<max_tool_calls: int64, max_turns_after_tools: int64, wall_clock_seconds: double, temperature: (... 7 chars omitted)
child 0, max_tool_calls: int64
child 1, max_turns_after_tools: int64
child 2, wall_clock_seconds: double
child 3, temperature: int64
status: string
error: null
tool_calls_made: int64
latency_ms: int64
prefetch: list<item: struct<id: string, name: string, args: struct<date: timestamp[s], days: int64>, result: s (... 1566 chars omitted)
child 0, item: struct<id: string, name: string, args: struct<date: timestamp[s], days: int64>, result: struct<avail (... 1554 chars omitted)
child 0, id: string
child 1, name: string
child 2, args: struct<date: timestamp[s], days: int64>
child 0, date: timestamp[s]
child 1, days: int64
child 3, result: struct<available: bool, date: timestamp[s], reason: string, run_date: timestamp[s], stored_max_game_ (... 1465 chars omitted)
child 0, available: bool
child 1, date: timestamp[s]
child 2, reason: string
child 3, run_date: timestamp[s]
child 4, stored_max_game_date: timestamp[s]
child 5, stored_rows: int64
child 6, stored_seasons: list<item: string>
child 0, item: string
child 7, days_stale_stored: int64
child 8, dump_present: bool
child 9, dump_max_game_date: null
...
ld 1, dropped: int64
summary: string
findings: list<item: struct<kind: string, severity: string, evidence: struct<tool: string, args: struct<date: (... 343 chars omitted)
child 0, item: struct<kind: string, severity: string, evidence: struct<tool: string, args: struct<date: timestamp[s (... 331 chars omitted)
child 0, kind: string
child 1, severity: string
child 2, evidence: struct<tool: string, args: struct<date: timestamp[s], days: int64>, values: struct<available: bool, (... 267 chars omitted)
child 0, tool: string
child 1, args: struct<date: timestamp[s], days: int64>
child 0, date: timestamp[s]
child 1, days: int64
child 2, values: struct<available: bool, reason: string, days_stale: int64, mae_pts: double, mae_reb: double, mae_ast (... 190 chars omitted)
child 0, available: bool
child 1, reason: string
child 2, days_stale: int64
child 3, mae_pts: double
child 4, mae_reb: double
child 5, mae_ast: double
child 6, largest_residual_pts: double
child 7, model_pts: double
child 8, baseline_pts: double
child 9, model_reb: double
child 10, baseline_reb: double
child 11, model_ast: double
child 12, baseline_ast: double
child 13, known_missing_count: int64
child 3, text: string
generated_at: timestamp[s]
to
{'date': Value('timestamp[s]'), 'run_date': Value('timestamp[s]'), 'status': Value('string'), 'summary': Value('string'), 'findings': List({'kind': Value('string'), 'severity': Value('string'), 'evidence': {'tool': Value('string'), 'args': {'date': Value('timestamp[s]'), 'days': Value('int64')}, 'values': {'available': Value('bool'), 'reason': Value('string'), 'days_stale': Value('int64'), 'mae_pts': Value('float64'), 'mae_reb': Value('float64'), 'mae_ast': Value('float64'), 'largest_residual_pts': Value('float64'), 'model_pts': Value('float64'), 'baseline_pts': Value('float64'), 'model_reb': Value('float64'), 'baseline_reb': Value('float64'), 'model_ast': Value('float64'), 'baseline_ast': Value('float64'), 'known_missing_count': Value('int64')}}, 'text': Value('string')}), 'tool_calls_made': Value('int64'), 'model_id': Value('string'), 'latency_ms': Value('int64'), 'generated_at': Value('timestamp[s]'), 'grounding': {'checked': Value('int64'), 'dropped': Value('int64')}}
because column names don't matchNeed help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
NBA Player Game Logs
One row per player per regular-season game, 2021-22 through 2025-26, in a canonical schema used by cbratkovics/nba-ai-ml for point-in-time feature engineering and batch prediction of points, rebounds, and assists.
Provenance
- Historical backfill: Eoin Moore, Historical NBA Data and Player Box Scores, Kaggle (
eoinamoore/historical-nba-data-and-player-box-scores), version 515, CC0 1.0. OnlyPlayerStatistics.csvandTeamHistories.csvare used; rows carrysource = kaggle_v515. - Daily updates: the same Kaggle dataset, re-downloaded by the nightly GitHub Actions job. Rows within 7 days of the newest stored game are reconciled against the stored rows; new or changed rows carry
source = kaggle_daily. There is no live collection from NBA.com: stats.nba.com is not reachable from GitHub Actions runners. - The
sourcecolumn records which pipeline produced each row.
Both sources are derived from NBA.com. This dataset is published for non-commercial, personal portfolio use; the CC0 label covers the upstream Kaggle packaging, not the NBA's underlying terms.
Schema
One row per (player_id, game_id). Columns: game_id, game_date, season,
player_id, player_name, team, opponent, home, minutes, pts, reb, ast,
fgm, fga, fg3m, fg3a, ftm, fta, oreb, dreb, stl, blk, tov, pf,
plus_minus, source. Team columns are abbreviations. Regular season only.
Files
game_logs/game_logs_YYYY-YY.parquet, one per season.
| Season | Rows | Games | First game | Last game |
|---|---|---|---|---|
| 2021-22 | 25,826 | 1,230 | 2021-10-19 | 2022-04-10 |
| 2022-23 | 25,687 | 1,230 | 2022-10-18 | 2023-04-09 |
| 2023-24 | 26,095 | 1,230 | 2023-10-24 | 2024-04-14 |
| 2024-25 | 26,158 | 1,223 | 2024-10-22 | 2025-04-13 |
| 2025-26 | 26,648 | 1,230 | 2025-10-21 | 2026-04-12 |
Known limitations
- Playoffs, play-in, preseason, and All-Star games are excluded. NBA Cup (in-season tournament) group and knockout games are included and the Cup final is excluded, matching official regular-season accounting.
- Rows for players who did not play (DNP) are dropped: 29,585 rows (2021-22: 5,495; 2022-23: 5,855; 2023-24: 6,290; 2024-25: 6,334; 2025-26: 5,611). A row counts as DNP when minutes are missing or 0, or when the box-score comment is populated.
- Games missing from the dump: 7 regular-season games were postponed and never re-captured upstream, so their box scores are absent here and are not filled in from any other source. 2024-25: 0022400524 (2025-01-09, LAL vs CHA), 0022400532 (2025-01-11, ATL vs HOU), 0022400537 (2025-01-11, LAL vs SAN), 0022400538 (2025-01-11, LAC vs CHA), 0022400617 (2025-01-22, NOP vs MIL), 0022400627 (2025-01-23, UTA vs WAS), 0022400988 (2025-03-17, SAN vs ORL).
- Team abbreviations follow
TeamHistories.csvfor the season in which the game was played.
- Downloads last month
- 7,695
Models trained or fine-tuned on cbratkovics/nba-game-logs