Uber ATG's self-driving vehicles measure a multitude of possible scenario variations to answer the age-old question: "how does the pedestrian cross the road?"
Uber employs statistical modeling to find anomalies in data and continually monitor data quality.
We built a backtesting service to better assess financial forecast model error rates, facilitating improved forecast performance and decision making.
In October 2019, Uber hosted our second annual Moving The World With Data meetup, showcasing some of our most interesting data science challenges in 2019.
We implemented a Kappa architecture at Uber to effectively backfill streaming data at scale, ensuring accurate data in our platform.
We engineered full SQL support on Apache Pinot to enable quick analysis and reporting on aggregated data, leading to improved experiences on our platform.
As part of Uber Visualization's all-team hackathon, we built SpeedsUp, a project using machine learning to process average speeds across a city, cluster the results, and overlay them on a street map.
In 2019, Uber's Data Platform team leveraged data science to improve the efficiency of our infrastructure, enabling us to compute optimum datastore and hardware usage.
We share technical challenges and lessons learned while productionizing and scaling XGBoost to train distributed gradient boosted algorithms at Uber.
As part of Uber Visualization's all-team hackathon, we built Urban Symphony, an Uber Movement visualization that adds an audio component to traffic speed patterns.
With the release of deck.gl version 7.3, Uber’s open source visualization tool now supports rendering massive geospatial data sets formatted according to the OGC 3D Tiles community standard.
Uber has embraced Presto, a high performance, distributed SQL query engine, and joined the Presto Foundation. Meet the Uber engineers who contribute to and use Presto on a daily basis.
Suzette Puente, Uber Data Science Manager, shares how she applies her graduate work in statistics to forecast traffic patterns and generate better routes.
Data science helps Uber determine which tables in a database should be off-boarded to another source to maximize the efficiency of our data warehouse.
Dawn Woodard, Director of Data Science, considers travel time prediction one of Uber's most interesting mapping problems.
Uber Principal Engineer Waleed Kadous discusses how we assess technologies our teams can leverage to improve the reliability and performance of our platform.
Uber Director of Data Science Franziska Bell discusses how we created data science platforms at Uber, letting employees of all technical skills perform forecasts and analyze data.
Uber engineers created uSCS, a Spark-as-a-Service solution that helps manage Apache Spark jobs throughout large organizations.
In a selection of presentations delivered at a June 2019 Uber meetup, we discuss how to use H3, our open source hexagonal indexing system, to facilitate the granular mining of large geospatial data sets.
Uber's Marketplace simulation platform leverages ML to rapidly prototype and test new product features and hypotheses in a risk-free environment.