## UmapExprNamespace


UMAP and nearest neighbor search on an expression.


Usage

``` python
UmapExprNamespace(expr)
```


Available as `pl.col(...).umap` after `import polars_umap`.


## Methods

| Name | Description |
|----|----|
| [fit_transform()](#fit_transform) | Embed this column with UMAP. Takes the same parameters as [polars_umap.umap()](umap.md#polars_umap.umap). |
| [from_knn()](#from_knn) | UMAP on a precomputed kNN graph, with this expression as the indices. |
| [knn()](#knn) | Approximate k nearest neighbors. Takes the same parameters as [polars_umap.knn()](knn.md#polars_umap.knn). |

------------------------------------------------------------------------


#### fit_transform()


Embed this column with UMAP. Takes the same parameters as [polars_umap.umap()](umap.md#polars_umap.umap).


Usage

``` python
fit_transform(**params)
```


##### Examples

``` python
>>> import polars as pl
>>> import polars_umap  # noqa: F401
>>> df = pl.DataFrame({"v": [[float(i), float(i % 5), float(i % 3)] for i in range(50)]})
>>> out = df.select(pl.col("v").umap.fit_transform(random_state=0))
>>> out.schema["v"] == pl.Array(pl.Float32, 2)
```

True

------------------------------------------------------------------------


#### from_knn()


UMAP on a precomputed kNN graph, with this expression as the indices.


Usage

``` python
from_knn(distances, **params)
```


See [polars_umap.umap_from_knn()](umap_from_knn.md#polars_umap.umap_from_knn).


------------------------------------------------------------------------


#### knn()


Approximate k nearest neighbors. Takes the same parameters as [polars_umap.knn()](knn.md#polars_umap.knn).


Usage

``` python
knn(**params)
```
