The axia package provides an implementation of an extension of the Shifted-Beta-Geometric model by P. Fader & B. Hardie [1] to individuals in contractual settings with multiple predictor variables.

Axia offers a survival analysis like approach to modeling the behaviour of individuals in contractual settings. Given a set of features, ages and status (alive or dead), this package builds on top of the model developed by P. Fader & B. Hardie [1], to infer hazard curves, retention curves and discounted-life-time-value.

Head over to the examples folder to learn how to use axia.

Installation

pip3 install axia

Dependencies

• numpy~=1.21.3
• scipy~=1.7.1
• pandas~=1.3.4

References

1. "HOW TO PROJECT CUSTOMER RETENTION", P. Fader & B. Hardie
2. "Customer-Base Valudation in a Contractual Setting: The Perils of Ignoring Heterogeneity", P. Fader & B. Hardie