Unsupervised Models

v0.1.7

Clustering and dimensionality reduction: KMeans with elbow and silhouette analysis, DBSCAN, hierarchical clustering, Gaussian mixtures, PCA with explained variance and TSNE.

Quick Start

cluster.py

python

import synaptix as sx

df = sx.load_dataset("iris")
X = df.drop(columns="species")

# KMeans with built-in diagnostics
km = sx.unsupervised.KMeans(n_clusters=3, random_state=42)
km.fit(X)
print("Inertia:", km.inertia_)

metrics = sx.metrics.clustering_metrics(X, km.labels_)
print(metrics)   # silhouette, Calinski-Harabasz, Davies-Bouldin

km.plot_elbow(max_k=10)   # elbow curve renders inline

API Reference

KMeans

KMeans(n_clusters=8, random_state=None)

DBSCAN

DBSCAN(eps=0.5, min_samples=5)

HierarchicalClustering

HierarchicalClustering(n_clusters=3, linkage='ward')

GaussianMixture

GaussianMixture(n_components=3)

PCA

PCA(n_components=None)

transformed DataFrame

TSNE

TSNE(n_components=2, perplexity=30)