ClusterAudienceKit

Segment millions of customers in <1 second. Know who matters.

RFM analysis, customer lifetime value prediction, churn detection, and lookalike modeling—all in one production-ready package. Process 1M+ customers instantly.

30-Second Start

from clusteraudiencekit import Segmentation, CLV

# Segment your customers
segments = Segmentation(df).fit()
print(segments.summary)  # Automatic RFM + clustering

# Find high-value customers
clv = CLV(df)
vips = clv.top_customers(n=100)
print(f"Top 100 worth: ${clv.total_value(vips):,.0f}")

Why ClusterAudienceKit?

The Problem:

• Marketing teams manually segment (outdated, slow)
• No clear view of customer value
• Churn prediction requires multiple tools
• Building lookalike audiences is complex

The Solution:

• Automatic RFM segmentation (no configuration)
• Customer lifetime value prediction
• Churn scoring and early warning
• Lookalike audience generation
• Sub-second processing (1M+ customers)

Quick Start

from clusteraudiencekit import Segmentation, CLV

# Load customer data
df = load_customers()

# Automatic segmentation
segments = Segmentation(df).fit()
print(segments.summary)

# Calculate lifetime value
clv = CLV(df)
high_value = clv.top_customers(n=100)

Process 1M+ Customers in <1s

Optimized clustering on millions of records:

• K-means, DBSCAN, Hierarchical clustering, Gaussian Mixture Models
• Parallel processing
• GPU acceleration support

Key Features

• RFM analysis built-in
• 4 clustering algorithms
• Customer Lifetime Value (CLV)
• Churn prediction
• Lookalike modeling
• Neural network clustering
• Sub-second processing

Key Features

Audience Analysis

• RFM segmentation (Recency, Frequency, Monetary)
• 4 clustering algorithms (K-means, DBSCAN, Hierarchical, Gaussian Mixture Model). Spectral clustering and Isolation Forest are not implemented anywhere in this codebase (not even as stubs) despite being listed here previously — removed until they're real.
• Customer Lifetime Value (CLV) prediction
• Churn detection and scoring
• Lookalike audience modeling

Scalability

• Process 1M+ customers in <1 second
• Rust-powered core for speed
• Streaming data support
• Batch and real-time APIs

Integration

• Pandas/Polars DataFrames
• Cloud storage (S3, GCS)
• SQL databases
• BI tools (Tableau, Looker)

Requirements

• Python 3.10+
• NumPy ≥1.20.0
• Scikit-learn ≥1.0.0
• Pandas ≥1.3.0
• Rust core (precompiled wheels)
• Optional: Polars ≥0.18.0

Installation

pip install clusteraudiencekit
# or with uv
uv pip install clusteraudiencekit

# Verify installation
clusteraudiencekit --version

Use Cases

• Customer segmentation
• Churn prediction
• Retention campaigns
• CLV analysis
• Lookalike targeting
• Campaign personalization

Examples

from clusteraudiencekit import Segmentation, Churn, Lookalikes

# Segment customers
seg = Segmentation(df)
high_value = seg.segments['high_value']

# Predict churn
churn = Churn(df)
at_risk = churn.predict_churn(threshold=0.7)

# Find lookalikes
similar = Lookalikes(df).find_similar(seed_customers)

Benchmarks

Documentation

• API Reference
• Examples
• Benchmarks

License

MIT License - See LICENSE