Case Study: Vision Banco cuts model build time and doubles offer take-up with H2O.ai Driverless AI

Visión Banco cuts model build time 96% with H2O.ai

Vision Banco, a leading bank in Paraguay's microfinance segment, faced a significant challenge in building credit risk models. Their traditional methods, including an IBM SPSS model that took a year to build and open source PMML deployment, were slow and effort-heavy. They needed to assess credit risk faster and more accurately to expand their services and move into predictive analytics. To address this, they turned to H2O.ai and implemented its Driverless AI platform.

The solution from H2O.ai automated the work with automatic feature engineering, model validation, and deployment via H2O's MOJO scoring pipelines. This cut the model build time from an average of six months down to less than a week. The more accurate models allowed Vision Banco to target offers much more effectively, which doubled the number of customers accepting offers and purchasing credit products. This sharper targeting is estimated to be worth millions in additional revenue for the bank.

Visión Banco