# Data card: `data/` This example uses **no curated data**: `code/example_challenge_notebook.ipynb` reads the `annual_indices` table of the challenge dataset directly from the Hub and prepares it in memory. ## Files None. ## Source read by the code | Source | Exact version | How the code reads it | License | Years used | Preparation (in the notebook) | Used for | |---|---|---|---|---|---|---| | [Challenge dataset](https://huggingface.co/datasets/CMU-OSPO/protostar-climate-train-to-2018), `annual_indices` | latest revision of the dataset repository | `pd.read_parquet("hf://datasets/CMU-OSPO/protostar-climate-train-to-2018/annual_indices")` | see the dataset card | up to 2018 | long format; a year is valid if the station missed at most 10 days (355-day rule) | Forecast input | ## Data rule check - [x] Every forecast input is weather data up to 2018-12-31. - [x] No 2019–2025 data is used.