Capital One on H2O Machine Learning
- H2O-3
Overview
In this customer testimonial, Capital One data scientist Brandon Herer explains why the company adopted H2O for modeling on large-scale data. He describes how H2O let Capital One iterate on terabyte-scale datasets much faster, avoid vendor lock-in through open source, and use H2O Flow to communicate modeling results clearly to executives and teams across the company.
Key takeaways
- Why Capital One needed a tool that could model terabyte-scale data quickly and iteratively
- How open source access reduced vendor lock-in risk and let Capital One extend H2O when needed
- How in-memory processing changed model run times from "run a model and go to lunch" to near-immediate results
- How H2O Flow helped Capital One communicate modeling work to executives and less technical stakeholders
- Why Capital One valued H2O's transparency, including the ability to inspect the source code behind tree and network models
- How Capital One used H2O across its data science org, including hosting regular H2O community meetups
Start owning your AI