add birch clustering

src/depiction/tools/clustering.py

@@ -5,7 +5,7 @@
 import numpy as np
 import xarray
 from numpy.typing import NDArray
-from sklearn.cluster import KMeans, BisectingKMeans
+from sklearn.cluster import KMeans, BisectingKMeans, Birch
 
 from depiction.clustering.extrapolate import extrapolate_labels
 from depiction.clustering.maxmin_sampling import maxmin_sampling
@@ -19,6 +19,7 @@
 class MethodEnum(Enum):
     KMEANS = "kmeans"
     BISECTINGKMEANS = "bisectingkmeans"
+    BIRCH = "birch"
 
 
 app = cyclopts.App()
@@ -55,7 +56,7 @@ def clustering(
     method_params: str,
     n_best_features: int = 30,
     n_samples_cluster: int = 10000,
-    n_landmarks: int = 50,
+    n_landmarks: int = 200,
 ) -> None:
     rng = np.random.default_rng(42)
 
@@ -116,6 +117,9 @@ def compute_labels(features: NDArray[float], method: MethodEnum, method_params:
     elif method == MethodEnum.BISECTINGKMEANS:
         clu = BisectingKMeans(n_clusters=10).fit(features)
         return clu.labels_
+    elif method == MethodEnum.BIRCH:
+        clu = Birch(n_clusters=10).fit(features)
+        return clu.labels_
     else:
         raise ValueError(f"Method {method} not implemented")
 

src/depiction_cluster_sandbox/run_cluster_sandbox.py

@@ -15,18 +15,20 @@ def main():
 
     samples = []
     samples += ["concatenated"]
-    cluster_variants = [
-        "cluster_kmeans_default.png",
-        "cluster_bisectingkmeans_default.png",
-        "cluster_kmeans_featscv.png",
-        "cluster_bisectingkmeans_featscv.png",
-        "cluster-umap_kmeans_default.png",
-        "cluster-umap_kmeans_featscv.png",
-    ]
-    result_files = [
-        work_dir / "work" / sample / cluster_variant for sample in samples for cluster_variant in cluster_variants
+
+    cluster_algos = ["kmeans", "bisectingkmeans", "birch"]
+    cluster_artifacts = [
+        file
+        for algo in cluster_algos
+        for file in [
+            f"cluster_{algo}_default.png",
+            f"cluster-umap2-{algo}_default-cluster.png",
+            f"cluster-umap2-{algo}_default-image_index.png",
+        ]
     ]
 
+    result_files = [work_dir / "work" / sample / artifact for sample in samples for artifact in cluster_artifacts]
+
     snakemake = SnakemakeInvoke(continue_on_error=False, snakefile_name=snakefile_path)
     snakemake.invoke(
         work_dir=work_dir,