Predicting Taxi Fare in New York

Today, the benefit of Machine Learning is conditioned to its deployment in real-time. In this talk, Bart will explain how to deploy a real-time taxi fare prediction engine to power an Uber-like application. Along the cycle of developing such project, we will highlight key lessons we learned: - Understand the problem before building models - Do not add features for the sake of features - Try as many algorithms as possible - Simplify your pipeline before deployment Bart Koek, Solution Architect at Dataiku: Bart is in charge of helping companies in the Nordics region and across Europe to address, implement, and deploy data-driven projects. Dataiku is a collaborative data science platform, which integrates all the capabilities required to build end-to-end highly specific services that turn raw data into business-impacting predictions quickly.