Survival Analysis in Modeling Risk, Churn & More

Hi LAML Member,

We have another great speaker, Goutham Balaraman, coming to talk about Machine Learning in production. Please come join us at Zefr to learn about how to survival analysis.

Here is the schedule:

5:30 Meetup attendees arrive
5:30 - 6:30 Food, drinks, and mingling
6:30 - 7:15 Presentation
7:15 - 7:30 Q&A
8 PM Doors close

Abstract

When predicting customer churn or loan defaults, standard ML models usually ask a binary question: "Will this happen within a year, yes or no?" But treating an event in Month 2 exactly like an event in Month 11 throws away your most valuable feature: time. This talk explores the paradigm shift from traditional classification to Survival Analysis. We’ll walk through the architecture needed to stop predicting "if" and start predicting "when." Drawing on real-world experience in credit risk modeling, we’ll walk through the use cases where survival analysis works better than traditional classifiers.

Speaker Bio

Goutham Balaraman is the Chief Technology Officer at Crossroads Equipment Lease & Finance, where he directs data, analytics, and engineering initiatives. With nearly two decades of experience in FinTech, Goutham has spent his career building the infrastructure behind modern financial services. Before Crossroads, he served as CIO at Cartiga and held senior technology and analytics leadership roles at loanDepot, Numerix, and Intercontinental Exchange (ICE). A strong advocate for open-source and hands-on engineering, he is also the co-author of the QuantLib Python Cookbook. He holds a Ph.D. in Physics from Georgia Tech and lives in Los Angeles with his wife, two kids, and two cats.