Hacker News new | past | comments | ask | show | jobs | submit login

I noticed that this (density-based clustering) is conceptually very different from an earlier Netflix approach (RAD - http://techblog.netflix.com/2015/02/rad-outlier-detection-on...) which performs RPCA and estimates thresholds. Are both systems in place? In which cases would one system outperform another?

RAD requires at least 2 cycles of data, whereas the real-time analytics team tends to operate on very short time frames. Our approach is less accurate but able to make quick decisions, which is the objective in this case.

Both systems are certainly in place, though they serve different business use cases. RAD isn't used in real-time for operational decisions and Kepler isn't used for anything that Data Science uses RAD for.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact