README.md

---
license: mit
language:
- en
task_categories:
- time-series-forecasting
- tabular-regression
size_categories:
- 10K<n<100K
tags:
- tsfile
- timeseries
- modality:timeseries
- format:tsfile
- finance
- stock-market
modality: timeseries
pretty_name: S&P 500 Index History (TsFile format)
configs:
- config_name: default
  data_files:
  - split: train
    path: "spx.tsfile"
---

# S&P 500 Index History (TsFile format)

This dataset contains the historical S&P 500 index observations published in
the source repository. Each dated row provides open, high, low, close,
adjusted-close, and volume values for time-series forecasting and tabular
analysis.

Modalities: Time-series

## Source and scale

- Original dataset: [misikoff/SPX](https://huggingface.co/datasets/misikoff/SPX)
- Source revision: `498783fd0df5ad1f13c98e42329c60b8f8f9c6c1`
- Source file: `^SPX.csv` (CSV)
- Converted scale: 24,167 observations in one TsFile table.
- Observed date range: 1928-01-03 through 2024-03-15.
- Source split: one table; no validation or test split is defined.

## TsFile schema

| Column | Role | TsFile type | Source meaning |
|---|---|---|---|
| Time | TIME | INT64 (ms) | Source `Date`, parsed at UTC midnight |
| open | FIELD | DOUBLE | Opening index value |
| high | FIELD | DOUBLE | Highest index value |
| low | FIELD | DOUBLE | Lowest index value |
| close | FIELD | DOUBLE | Closing index value |
| adj_close | FIELD | DOUBLE | Adjusted closing index value |
| volume | FIELD | INT64 | Source volume value |

There are no TAG columns because the source contains one index series.

## Conversion notes

- `Date` is represented losslessly as integer-millisecond `Time`; the
  redundant source date text is not duplicated as a FIELD.
- Source measurement names are normalized to lowercase TsFile identifiers;
  `Adj Close` becomes `adj_close`.
- All 24,167 rows and all six measurements are retained. The conversion does
  not infer a market-data provider or provenance beyond the source card.

## Files and usage

- `spx.tsfile`

```python
from tsfile import TsFileReader

with TsFileReader("spx.tsfile") as reader:
    table = next(iter(reader.get_all_table_schemas()))
    with reader.query_table(table, ["close", "volume"], batch_size=1024) as result:
        batch = result.read_arrow_batch()
        if batch is not None:
            print(batch.to_pandas().head())
```

## License

The source dataset is licensed under the MIT License.

## Usage

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

```python
from pathlib import Path
from tsfile import TsFileReader

path = Path("spx.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())
```