Live data from Hacker News

Amundsen – Lyft’s data discovery and metadata engine

eng.lyft.com

11–13 of 13 posts

Re: Amundsen – Lyft’s data discovery and metadata engine

#11
post #2

We did a presentation in Strata SF 2019 last week which detail could be found at https://conferences.oreilly.com/strata/strata-ca/public/sche... . The slide could be found at https://www.slideshare.net/taofung/strata-sf-amundsen-presen... .

How do you deal with discoverability of map keys and struct fields? That's something we just added after such datatypes became more prominent in our DWH.

Re: Amundsen – Lyft’s data discovery and metadata engine

#13

Would someone mind explaining how ML models are used in the real world? In this case Lyft has users, trips, points of interest (destinations). What would they create a model for? How would it improve driver/rider experience? I've gone through a few TensorFlow tutorials, but still can't grok what the appropriate use-case is. Edit: I really appreciate the responses @tedsanders and @theossuary - thank you so much! Not s…

I received an offer from Lyft last year to do data science. Here are some potential use cases of machine learning for a taxi company: -Predicting wait times -Predicting the best route -Predicting where to send drivers before requests come in -Predicting demand so that you can surge price predictively -Predicting what assets/requests to prefetch to the phone and when -Predicting when drivers will churn and what tactic…

Predicting age, gender, and cost of the route and timings for en route, and doing shopping online for groceries and beverages in the cab, and pick them on their return to home.
Post reply on HN