Clean Data is Great Data

Byron North
Saturday, 2:00pm – 5 min
Marc Farra
Development Seed

Machine learning can be used to predict map features with high levels of accuracy. However, the outputs of the process are not usually suitable for direct import into OpenStreetMap because of their format, noise in the data, or duplication of existing OpenStreetMap features. We’ll present a variety of technical workflows and tools for solving these problems. We’ll also show how these approaches can be abstracted for non-machine learning use cases like better bulk uploads and data conflation.