outputs/cluster_optimization_log.csv

algorithm,umap_n_neighbors,umap_n_components,umap_metric,hdbscan_min_cluster_size,hdbscan_min_samples,hdbscan_metric,n_clusters,noise_ratio,min_size,max_size,too_small,too_large,silhouette_cosine,score,optimizer_recommendation
UMAP + HDBSCAN (external),10,5,cosine,8,,euclidean,23,0.0,6,100,0,0,0.082,-0.082,Keep this parameter set; it satisfies the 15-25 cluster target best.
UMAP + HDBSCAN (external),10,5,cosine,8,3.0,euclidean,32,0.0,8,74,0,0,0.0664,20.9336,Raise HDBSCAN min_cluster_size or increase UMAP n_neighbors to merge nearby themes.
UMAP + HDBSCAN (external),10,5,cosine,5,,euclidean,35,0.0,7,72,0,0,0.0695,29.9305,Raise HDBSCAN min_cluster_size or increase UMAP n_neighbors to merge nearby themes.
UMAP + HDBSCAN (external),10,5,cosine,8,1.0,euclidean,39,0.0,8,73,0,0,0.0574,41.9426,Raise HDBSCAN min_cluster_size or increase UMAP n_neighbors to merge nearby themes.
UMAP + HDBSCAN (external),10,5,cosine,5,3.0,euclidean,53,0.0,5,73,0,0,0.0619,83.9381,Raise HDBSCAN min_cluster_size or increase UMAP n_neighbors to merge nearby themes.
UMAP + HDBSCAN (external),10,5,cosine,5,1.0,euclidean,75,0.0,5,29,0,0,0.0639,149.9361,Raise HDBSCAN min_cluster_size or increase UMAP n_neighbors to merge nearby themes.