Data Science, Delivered Continuously

Explore a 48-minute conference talk from GOTO Berlin 2017 featuring Christian Deger and Arif Wider as they discuss continuous delivery in data science. Learn how AutoScout24, Europe's largest online car marketplace, leverages its vast data on market prices to create a live price estimation service for used vehicles. Discover the implementation of a Random Forest prediction model and the challenges faced in developing a continuous delivery pipeline for machine learning models. Gain insights into model validation workflows, technical lessons learned, and solutions to issues such as warmup time. Understand how AutoScout24 empowers its users with data-driven decision-making tools for buying and selling cars. Perfect for those interested in practical applications of data science, machine learning, and continuous delivery in a real-world business context.

Introduction
About Arif Wider
Random Forest
H2O
Continuous Delivery Pipeline
Model Validation Workflow
What we learned
Technical challenges
Warmup time
Technical lessons
Conclusions
Questions