Spyros Sakellariadis, Joshua Peschel. To collect the right data for a complex machine learning experiment might take months, as you plan out choices and placements of sensors, data schemas, data logging intervals, and so on to get optimal data sets for running analytical models that produce the insights you need. In this session we'll cover the planning architecture and end-to-end experiments for measuring and analyzing hydrologic data (e.g., soil moisture) for two ongoing operations in the greater Chicagoland area: (1) determining water balances for agricultural lands; and (2) predicting and preventing urban flooding. This talk will elucidate the complexities and provide recommendations and best practices for working with environmental data in the wild. Go to https://channel9.msdn.com/ to find the recording of this session.