---
id: lifetimes
version: "0.11.3"
license: MIT
license_treatment: permissive
maintenance: abandoned
---

# Lifetimes — Measure customer lifetime value in Python

License: permissive · Maintenance: abandoned · Downloads: 255.2K/mo

## What it is and what it does

Lifetimes is a Python library for modeling and predicting customer lifetime value (CLV) and churn using probabilistic methods grounded in survival analysis. It implements models like BG/NBD and Pareto/NBD that estimate how often a customer will return and their long-term value based on historical transaction data (frequency, recency, and age). The library treats customer engagement as a "birth-death" process: users are alive while active and may die (churn) after inactivity.

The package depends on numpy, scipy, pandas, autograd, and dill for numerical computation and serialization. It is designed for business analytics workflows where you have transaction histories and want to segment customers by predicted value or identify likely churners. The library was last released in 2020 and its repository is now archived, meaning no active maintenance or updates are occurring.

Use it for:

- Predict which customers are likely to churn based on their purchase history and engagement patterns.
- Estimate the lifetime value of customers to prioritize retention and marketing spend.
- Forecast repeat purchase frequency and timing for subscription or e-commerce businesses.
- Segment customers into cohorts by predicted value for targeted retention campaigns.
- Analyze user engagement on websites or apps to identify inactive or at-risk users.

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

Lifetimes models customer behavior and predicts lifetime value, churn, and repeat purchase likelihood using probabilistic survival-analysis methods on transaction history.

Yes, if you need CLV modeling and can accept an unmaintained package. The library is stable for its core use case (BG/NBD and Pareto/NBD models) and has no known vulnerabilities. However, verify dependency compatibility with your Python version first, and consider whether a maintained fork or alternative exists for your specific use case, given the project's archived status since 2024.

## Install

pip install lifetimes
uv add lifetimes
poetry add lifetimes

## Installing Lifetimes

Before you install: Low install friction; pure Python wheel. Repository is archived and last release was 2020-07-06, over 4 years old. Maintenance is abandoned, so no active bug fixes or dependency updates.

License in practice: MIT license permits commercial and private use with minimal restrictions; you must retain the license notice in distributions.

Quickstart:

pip install lifetimes

from lifetimes import BetaGeoFitter
import pandas as pd

bgf = BetaGeoFitter()
bgf.fit(data)  # data: DataFrame with frequency, recency, T columns

Requires numpy, scipy, pandas, autograd, and dill; no Python version constraint specified in package metadata.

Verify before relying:

- Whether the package works reliably with modern Python versions (3.10+) given its last release predates widespread adoption.
- Current compatibility of dependencies (numpy, scipy, pandas, autograd, dill) with the pinned or unpinned versions Lifetimes expects.
- Whether community forks or maintained alternatives have superseded this archived project.

## Package facts

- License: MIT (permissive)
- Python support: unspecified
- Install friction: low
- Maintenance: abandoned
- Downloads: 255.2K/month (top 15,000 on PyPI)
- Known vulnerabilities: none known

## Tags

customer lifetime value prediction, churn prediction from user history, repeat purchase forecasting, BG/NBD model implementation, customer behavior analysis, survival analysis for business, CLV calculation library, customer-analytics, survival-analysis, churn-prediction

[View on SkillFed](https://skillfed.io/packages/lifetimes) · [View on PyPI](https://pypi.org/project/lifetimes/)