Datasets:

THULab
/

curry_data

pretty_name: Stephen Curry Game Log (TsFile)
modality: timeseries
authors: jawwaadsabree
task_categories:
  - time-series-forecasting
size_categories:
  - 1K<n<10K
tags:
  - tsfile
  - timeseries
  - modality:timeseries
  - format:tsfile
  - NBA
configs:
  - config_name: default
    data_files:
      - split: train
        path: curry_data.tsfile

Stephen Curry Game Log (TsFile)

This dataset is an Apache TsFile conversion of jawwaadsabree/CurryData.

Modalities: Time-series.

Overview

Per-game NBA stat lines for Stephen Curry across seasons 2009–2016.
Game statistics (PTS, REB, AST, MIN, FG/3PT/FT splits, etc.) plus engineered cyclic date features.
Each season is a device identified by the season TAG.
Converted observations: 1,204 rows across 1 TsFile file(s)
Source format: csv

TsFile schema

Time — source Date (epoch nanoseconds), converted to INT64 milliseconds.

Column	Role	Type	Meaning
`Time`	TIME	INT64 (ms)	sample timestamp
`season`	TAG	STRING	season (e.g. 2009_2010)
`Result`	FIELD	FLOAT	win/loss
`MIN`	FIELD	FLOAT	—
`REB`	FIELD	FLOAT	—
`AST`	FIELD	FLOAT	—
`BLK`	FIELD	FLOAT	—
`STL`	FIELD	FLOAT	—
`PF`	FIELD	FLOAT	—
`TO`	FIELD	FLOAT	—
`PTS`	FIELD	FLOAT	points
`FG_Made`	FIELD	FLOAT	—
`FG_Attempts`	FIELD	FLOAT	—
`c_3PT_Made`	FIELD	FLOAT	—
`c_3PT_Attempts`	FIELD	FLOAT	—
`FT_Made`	FIELD	FLOAT	—
`FT_Attempts`	FIELD	FLOAT	—
`Opponent_1`	FIELD	FLOAT	—
`Opponent_2`	FIELD	FLOAT	—
`Opponent_3`	FIELD	FLOAT	—
`Opponent_4`	FIELD	FLOAT	—
`Opponent_5`	FIELD	FLOAT	—
`Opponent_6`	FIELD	FLOAT	—
`Opponent_7`	FIELD	FLOAT	—
`Opponent_8`	FIELD	FLOAT	—
`Opponent_9`	FIELD	FLOAT	—
`Opponent_10`	FIELD	FLOAT	—
`Opponent_11`	FIELD	FLOAT	—
`Opponent_12`	FIELD	FLOAT	—
`Opponent_13`	FIELD	FLOAT	—
`Opponent_14`	FIELD	FLOAT	—
`Opponent_15`	FIELD	FLOAT	—
`Opponent_16`	FIELD	FLOAT	—
`Opponent_17`	FIELD	FLOAT	—
`Opponent_18`	FIELD	FLOAT	—
`Opponent_19`	FIELD	FLOAT	—
`Opponent_20`	FIELD	FLOAT	—
`Opponent_21`	FIELD	FLOAT	—
`Opponent_22`	FIELD	FLOAT	—
`Opponent_23`	FIELD	FLOAT	—
`Opponent_24`	FIELD	FLOAT	—
`Opponent_25`	FIELD	FLOAT	—
`Opponent_26`	FIELD	FLOAT	—
`Opponent_27`	FIELD	FLOAT	—
`Opponent_28`	FIELD	FLOAT	—
`Opponent_29`	FIELD	FLOAT	—
`Opponent_30`	FIELD	FLOAT	—
`Is_Home`	FIELD	FLOAT	—
`Year`	FIELD	FLOAT	—
`Month_Sin`	FIELD	FLOAT	—
`Month_Cos`	FIELD	FLOAT	—
`Day_Sin`	FIELD	FLOAT	—
`Day_Cos`	FIELD	FLOAT	—
`Day_of_Week_Sin`	FIELD	FLOAT	—
`Day_of_Week_Cos`	FIELD	FLOAT	—
`Day_of_Year_Sin`	FIELD	FLOAT	—
`Day_of_Year_Cos`	FIELD	FLOAT	—
`Is_Playoff`	FIELD	FLOAT	—
`Is_Regular_Season`	FIELD	FLOAT	—
`Is_Preseason`	FIELD	FLOAT	—

Conversion notes

season (from the source file name) is a TAG so each season is a separate device.
57 game-stat / engineered-feature columns kept as FLOAT/INT64; no columns dropped.

Source & license

Original dataset: https://huggingface.co/datasets/jawwaadsabree/CurryData
Author / publisher: jawwaadsabree
License: not declared by the original dataset; please defer to the original

Usage

Install the Apache TsFile Python SDK (pip install tsfile) and read a converted file:

from pathlib import Path
from tsfile import TsFileReader

path = Path("curry_data.tsfile")
with TsFileReader(str(path)) as reader:
    schemas = reader.get_all_table_schemas()
    print("tables:", list(schemas))
    table_name = next(iter(schemas))
    table = schemas[table_name]
    columns = [column.get_column_name() for column in table.get_columns()]
    print("columns:", columns)
    field_names = [
        column.get_column_name()
        for column in table.get_columns()
        if column.get_column_name() not in {"Time", "time"}
    ]
    if field_names:
        with reader.query_table(table_name, field_names[:3], batch_size=1024) as result:
            batch = result.read_arrow_batch()
            if batch is not None:
                print(batch.to_pandas().head())