> For the complete documentation index, see [llms.txt](https://docs.everesteer.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.everesteer.ai/data/feature-groups.md).

# Feature groups

How features are named, typed, binned and grouped into sets.

Every feature follows the pattern `feature_<Name>`: one capitalised codename, for example `feature_Tichka`. The codename is an obfuscated label; it carries no relationship to the source data. There is no group or number component in a feature name. Feature names are consistent across `train`, `validation` and `live`: the same name in one split is the same feature in the others.

| Property | Value                                                                                                             |
| -------- | ----------------------------------------------------------------------------------------------------------------- |
| Type     | `int8`                                                                                                            |
| Bins     | `0` to `9`, ten equal-count levels per exped                                                                      |
| Missing  | `-1`: the source was unavailable for that instrument and day. Treat it as missing, never as an ordinal below `0`. |

The `-1` convention applies per row: a feature can be present for most instruments on an exped and unavailable for a few. Handle it as NaN or as a distinct category in your model.

## Feature sets

Sets are named groups of features. Atlas publishes a single set, `all`. The live catalogue, with the member list of every set, is served by `eiq_features.json` and by `get_dataset_schema` on the API and MCP surfaces. Read it from there rather than from a static list.

To scout cheaply, subsample expeds or features yourself, then scale to `all` once a pipeline earns a promotion. The feature space is versioned with the dataset, so a version change can rename or re-bucket the members. Re-read the catalogue after every dataset update.

{% hint style="info" %}
Feature values are cross-sectional decile bins within each exped. A label like `9` does not mean "nine times bigger": it means "in the top tenth of that cross-section".
{% endhint %}


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:

```
GET https://docs.everesteer.ai/data/feature-groups.md?ask=<question>&goal=<endgoal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.