Apache Big Data Europe 2016

The collection, documentation, management and analysis of big data associated with non-motorized travel has not attracted enough attentions. This may not conform to the trend that cycling, walking and jogging are strongly advocated by governments to build low-carbon cities and also to improve peopleäó»s health conditions. This session will share the experience that quantify and characterize the non-motorized travel by means of tempo-spatial analysis. The data used in this case is captured from a famous online community for running amateurs sharing their activities. Around 0.5 million running and cycling records from 0.3 million people in Beijing are analyzed with machine learning and data science methodology in this case study. Spark ML with random forest algorithm, and grid search of the parameters selection have been used on the prediction upon weather, AQI and time.