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

# DataFrame

An OuterProduct`DataFrame` is a query against the tables in your workspace (defined below).  Importantly, a `DataFrame` is a lazy-query, meaning that it will not materialize data until it is needed for computation.  As a result, you can query across data at any scale as if it were in memory.

The `DataFrame` is built in Rust on [Apache DataFusion](https://datafusion.apache.org/), so that queries run as real relational plans against all registered data sources simultaneously. These can be files on S3, Snowflake, Databricks, and your local machine.

## Workspace

The OuterProduct `Workspace` provides one object to connect and query across multiple data sources.   Construct custom views of your data (e.g. filter, join, derive columns) for training models and reasoning over these models.

```python theme={null}
# Initialize your workspace
ws = op.Workspace()

# Chain operations off a registered table.
train = ws.table("customers").filter(op.col("signup_year") >= 2020)

# Or write SQL against the workspace.
train = ws.sql("SELECT * FROM customers WHERE signup_year >= 2020")
```