#CaseStudy – find out how #Airbnb boosts productivity & scalability by transforming raw data into features for training & inference.

Meet #Chronon - a declarative #FeatureEngineering framework: https://bit.ly/3OAUJNE

#InfoQ #AI #ML #Phyton #MLOps

Every ML Eng book and resource I've ever read recommends that any ML product should start incredibly small, starting with something that can be developed in a day or so. Then iterate on it, only pulling in new features and only updating the model whenever those prove to add predictive value.

Every ML project I've been on has started with "we know we need these 372 features and need 12-24 months to get an MVP".

#featureengineering #ml #mleng

Just published my recent article on my understanding of #FeatureEngineering

https://medium.com/@kamig4u/a-comprehensive-guide-to-feature-engineering-for-machine-learning-in-python-b017274129fe

Awesome sharing by Nikhil Simha on Chronon, Airbnb's feature engineering framework! We had > 80 people joining and ~30 great questions on Slido.

Thanks to Chip Huyen for hosting and Ammar Asmro running it behind the scenes!

RSVP for future meetups here: https://www.meetup.com/ml-meetups-virtual/

#featureengineering #meetup

Presenting #featureengineering and post-processing steps for improving FedCSIS 2022 #Challenge results: “Key Factors to Consider when Predicting the Costs of Forwarding Contracts” by QH Vu, L. Cen, D. Ruta, M. Liu. @FedCSIS
2022, ACSIS Vol. 30 p. 447–450; http://tinyurl.com/2c4jvd92

What if I told you the reason you weren’t getting your dream #mlops gig wasn’t because of leetcode?

Had this realization as I was putting together an article #featureengineering at scale for data scientists.

So much of the challenges are workflow based versus trying to optimize dicts & search.

Is there a central/curated list / site of #ML features? eg variations on dates, locations etc ? I'm thinking like this: https://towardsdatascience.com/machine-learning-with-datetime-feature-engineering-predicting-healthcare-appointment-no-shows-5e4ca3a85f96 #pandas #featureengineering