Case Study: Acordo Certo reduces consumer debt in Brazil with H2O.ai
Key results
The challenge
Acordo Certo, founded in Sao Paulo in 2015, is Brazil's largest online debt payment and re-negotiation platform, with 3.6 million Brazilians on the platform. Its mission to help people reduce debt and help creditors collect required a small data-science team to build accurate debt-repayment scoring models quickly.
The solution
Using H2O Driverless AI with automatic feature engineering and machine-learning interpretability, Acordo Certo's small team of data scientists built new debt-scoring models and reviewed results far faster than before.
“We have reduced the time to build and deploy models from about one week to a few days. This is massive for our team. We have also improved the predictive accuracy of the repayment model to over 76%.”
AMAndre MonteiroFounder and CTO, Acordo Certo
The results, in context
With H2O Driverless AI, Acordo Certo retires over $10M of debt per month, reduced model development time by 70%, and improved repayment-model predictive accuracy to nearly 80% (over 76%). Time to build and deploy models fell from about one week to a few days.