Andrea Pasqua

Good business decisions cannot be made with bad data. At Uber, we use aggregated and anonymized data to guide decision-making, ranging from financial planning to letting drivers know the best location for ride requests at a given time. But how can we ensure high quality for the data powering these decisions? Interruptions in the pipeline can introduce troublesome anomalies, such as missing rows or fields, and affect the data we rely on for future decisions.

Andrea Pasqua investigates the merits of using deep learning and other machine learning approaches in the area of forecasting and describes some of the machine learning approaches Uber uses to forecast time series of business relevance.

By Slawek Smyl, Jai Ranganathan, Andrea Pasqua Uber’s business depends on accurate forecasting. For instance, we use forecasting to predict the expected supply of drivers and demands of riders in the 600+ cities we operate in, to identify when our systems are having outages, to ensure we always have enough customer obsession agents managing our support systems, and of course, to plan our business expenditures.