Churn Probability algorithm needed based on historical data

Project Description

Baremetrics provides revenue-based financial metrics for startups (MRR, LTV, ARR, churn, etc).

We're working on a "Churn Probability" number to provide to our customers where we'd like to, as accurately as possible, give a % probability that a given customer of theirs will churn.

We have historical data for all of this, so my assumption is we'll need to use that data to develop an algorithm that can be tweaked based on when users have actually churned in the past.

This project happens to be in Ruby, so familiarity with that is a plus, but utlimately the dataset can be given to you in any format.