robworks-software/ccisd-teks-alignment-split

[!WARNING] Deprecated - use [`ccisd-teks-alignment`](https://huggingface.co/datasets/robworks-software/ccisd-teks-alignment) instead. This dataset is superseded: the two contain the same 428 rows with the same 12 columns; this copy only adds a train/validation/test partition, which you can reproduce in one line. Nothing here is unique to it. It stays online so existing references keep resolving, but it will not be updated. New work should point at robworks-software/ccisd-teks-alignment.

CCISD TEKS Alignment (pre-split)

The same 428 TEKS-to-course mappings as `ccisd-teks-alignment`, partitioned into train/validation/test.

If you are starting fresh, use `ccisd-teks-alignment` and split it yourself. This repo exists only to provide a fixed partition; it adds no rows and no fields.

Loading

from datasets import load_dataset

ds = load_dataset("robworks-software/ccisd-teks-alignment-split")

Splits

Limitations

Everything that applies to the base dataset applies here:

• —`instructional_units` is empty - all 299 training rows carry "Not specified".
• —Single district, high school only. CCISD's course sequencing for one year.
• —STAAR status drifts as Texas revises its assessment program.
• —The split is random, not stratified by course or subject. Subject distribution differs between splits, so per-subject evaluation on the 65-row test split rests on very small cell counts.
• —428 rows total is too small for meaningful train/test evaluation of most models.

License

CC0-1.0. TEKS standards are Texas Education Agency public records.

Citation

@dataset{ccisd_teks_alignment_split,
  title  = {CCISD TEKS Alignment (pre-split)},
  author = {Robworks Software},
  year   = {2025},
  publisher = {Hugging Face},
  url    = {https://huggingface.co/datasets/robworks-software/ccisd-teks-alignment-split}
}