UMAP/t-SNE/PCA scatter plot of work embeddings, coloured by cluster.
Usage
sm_plot_landscape(
corpus,
color_by = "cluster_id",
reducer = c("umap", "tsne", "pca"),
n_components = 2L,
dark = FALSE,
...
)Value
A ggplot object with one point per row of
corpus$embeddings, so works without an embedding are not shown.
Details
UMAP retains its Euclidean metric and automatic thread settings.
If its PCA dependency raises the known older-R NULL-input error, it
retries with base SVD initialisation and reports
scimapR_umap_pca_fallback. Other errors propagate unchanged.
Examples
corpus <- sm_example_corpus()
corpus <- sm_cluster_hdbscan(corpus, min_cluster_size = 10)
#> ✔ HDBSCAN clustering complete.
#> ℹ 5 clusters found, 0 noise points.
sm_plot_landscape(corpus)