> ## Documentation Index
> Fetch the complete documentation index at: https://ploti.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Dataset kinds

> Every dataset is a layer, a table, a document, a scanned map or a collection. The kind decides which calls read it, what you can download, and whether Data Feeds delivers it.

A dataset is one set of official records Ploti holds from one source. Each is one of five kinds, and a search names the kind of each match. The kind decides which calls read it, whether you can download it, and whether Data Feeds delivers it.

| Kind | What it is | Example | Read it with | Download | In Data Feeds |
| - | - | - | - | - | - |
| `layer` | Rows that each have a shape on the map | Orange County parcels, FEMA flood zones | [Row reads](https://ploti.ai/docs/developers/read-rows.md) | Yes, as GeoParquet | Yes |
| `table` | Rows without shapes | A city's building permit log | [Row reads](https://ploti.ai/docs/developers/read-rows.md) | Yes, as Parquet | Yes |
| `document` | One text, split into sections | A city's zoning ordinance | [Document reads](https://ploti.ai/docs/developers/documents.md) | Its text, and its original file | No |
| `scanned_map` | A map published as an image or PDF, with no text | A scanned future land use map | A question search | Its original file | No |
| `collection` | Many documents kept as one dataset | A town's planning board minutes | [Collection reads](https://ploti.ai/docs/developers/documents.md#collections) | One document at a time | Its table of documents, when it has one |

To search only some kinds, pass `kinds` to [`datasets search`](https://ploti.ai/docs/developers/find-datasets).

## Layers

A layer is a set of records that each have a shape: parcels, zoning districts, flood zones, roads. Each record is a row with columns, such as a parcel's owner, acreage and land use, plus its shape.

* **Understand it:** `datasets get` lists its columns, `datasets values` the codes a coded column takes, and `datasets profile` how full each column is.
* **Read it:** see [Read rows](https://ploti.ai/docs/developers/read-rows.md).
  * `datasets rows` reads rows with conditions.
  * `datasets aggregate` counts and totals them.
  * `datasets lookup` finds rows by parcel number, address or owner name.
  * `datasets related` finds what overlaps or lies near one parcel.
  * `datasets estimate` sizes a read first.
* **Download it:** `datasets download` writes its rows to a file, whole or narrowed. See [Download a dataset](https://ploti.ai/docs/developers/download.md).

## Tables

A table is rows without shapes, usually from a spreadsheet or database export: a permit log, a fee schedule, a list of tax delinquencies.

* **Read it:** the same calls as a layer. Its rows have no shapes, so a table refuses a `place` or a `bbox` (a rectangle on the map). Narrow it with a condition on its own columns instead, such as its city or parcel number column.
* **Download it:** `datasets download`, without a shape column.

## Documents

A document is one text Ploti has read and split into sections: a zoning ordinance, a comprehensive plan, a staff report.

* **Read it:** see [Documents and collections](https://ploti.ai/docs/developers/documents.md).
  * `documents search` finds passages by meaning or exact words.
  * `documents read` reads one section.
  * `documents outline` lists the sections and tables in order.
  * `documents tables list` and `documents tables read` read the tables found inside it, such as a district's dimensional standards.
* **Download it:** `documents download` writes the whole text to a file. `datasets files get` with `original` downloads the file Ploti read it from, such as the PDF.

## Scanned maps

A scanned map is a map a government published as an image or a PDF, not as data, such as a scanned future land use map or a zoning atlas page. Ploti keeps its title, a description and the area it covers, but it has no text to open.

* **Read it:** a `documents search` with a question can find it. It can't be opened with `documents read`.
* **Download it:** `datasets files get` with `original` downloads the image or PDF.

## Collections

A collection is many documents that belong together, kept as one dataset: a town's meeting minutes, an agency's yearly reports, a numbered series of resolutions. Each document in it has a **member key** (its id within the collection), a title and a date.

* **Read it:**
  * `collections documents list` lists the documents.
  * `collections documents search` finds words inside them and names the document and page of each match.
  * `collections documents read` opens one document by its member key.
  * A collection usually also has a table with one row per document. Read it like any other table.
* **Download it:** not as one file. Read the documents you need one at a time. Each comes with a link to its stored file.

```bash CLI theme={null}
ploti datasets search "zoning ordinance" --place "Tampa, FL" --kinds document
```

```python Python theme={null}
import ploti

ploti.datasets.search(["zoning ordinance"], place="Tampa, FL", kinds=["document"])
```

```bash REST theme={null}
curl -X POST https://api.ploti.ai/v1/datasets/search \
  -H "Authorization: Bearer $PLOTI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"queries": ["zoning ordinance"], "place": "Tampa, FL", "kinds": ["document"]}'
```

```json MCP theme={null}
{
  "tool": "datasets_search",
  "arguments": {
    "queries": ["zoning ordinance"],
    "place": "Tampa, FL",
    "kinds": ["document"]
  }
}
```


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.