This repository contains some experiments with analytics for cycling data.

See this blog post for some background.

To produce the kinds of maps in that post, place a bunch of TCX files with GPS and wattage data in a directory called activities and then run sbt console. Once the code is done compiling and you're at the REPL prompt, type this to generate a map:

com.freevariable.surlaplaque.GPSClusterApp.main(Array("-dactivities"))

By default, the map will go into a file in the current directory called slp.json. Currently all configuration is done through environment variables; some influential ones include:

• SLP_MMP_PERIOD defines the period, in seconds, to calculate mean maximal power for (defaults to 60 seconds)
• SLP_CLUSTERS defines the value of k for k-means clustering (defaults to 128)
• SLP_ITERATIONS defines the iteration count for clustering (defaults to 10)
• SLP_MASTER defines the Spark master to use (defaults to local[8])
• SLP_OUTPUT_FILE defines the name of the output file for GeoJSON data (defaults to slp.json)

There are other applications here too but their interfaces and functionality are rather more subject to change. More documentation is forthcoming.