autoaicluster 0.1.5

Enterprise-grade clustering solution with local web UI for preprocessing, clustering, and visualization.

Enterprise Cluster Solution

A Python package that launches a local web UI to build and run clustering pipelines with minimal code.

Installation

pip install autoaicluster

Quickstart (Web UI)

from autoai import AutoAI

# Binds to 0.0.0.0 by default; shows a localhost URL
model = AutoAI()
url = model.launch_ui()
print("UI available at:", url)

• To override host/port, use environment variables or constructor:
  • ECS_HOST or HOST (e.g., 0.0.0.0 for external access)
  • ECS_PORT or PORT

Python Workflow (No UI)

import pandas as pd
from enterprise_cluster_solution import run_pipeline, auto_segment

# dataframe or CSV path
df = pd.DataFrame({
    'age':[25,37,29,41,33,52,47,23,39,31],
    'income':[40,72,50,90,60,120,95,35,70,55],
    'city':['A','B','A','B','A','B','B','A','B','A']
})

# AutoML-style segmentation
result = auto_segment(df)
print('Algorithm:', result['algorithm'])
print('Silhouette:', result['silhouette'])
print('Counts:', result['label_counts'])

# Or manual pipeline
pipeline = run_pipeline(
    df,
    preprocessing={
        'imputation': {'numeric': 'mean', 'categorical': 'most_frequent', 'fill_value': 0},
        'encoding': 'onehot',
        'scaling': 'standard',
        'outliers': {'method': 'isoforest', 'contamination': 'auto'}
    },
    clustering={'algorithm': 'kmeans', 'params': {'n_clusters': 3}},
    visualization={'method': 'pca', 'n_components': 2}
)
print('KMeans silhouette:', pipeline['silhouette'])

Notes on access from containers/remote

• If running UI in a container or remote VM, bind to 0.0.0.0 and open http://<public-ip>:<port>.

Features

• Dtype editing; per-column and global imputation; include/exclude columns
• Outlier handling (IsolationForest), separate outliers artifact
• Algorithms: KMeans, K-Medoids, Agglomerative, BIRCH, DBSCAN, OPTICS, GMM
• Visualization: PCA, t-SNE, UMAP
• Metrics/insights: Silhouette, per-point silhouette, top features, cluster profiles
• Auto Segment: tries algorithms/params and selects best silhouette
• Artifacts: segmented.csv, outliers.csv

License

MIT