Given a log where each entry contains
- a timestamp,
- a url,
- a user identifier,
this app will analyse the user journeys and construct the graph of all journeys.
The result is two files
- a list of the unique nodes in the graph with number of visits
- a list of all the edges in the graph
These files are easily combined to produce a dot file, assuming that the resulting graph is small enough to be kept in memory.
To test the app on a laptop we need to
- build the app
mvn clean package
- generate a fake web log iwth 1000 user journeys.
scala src/main/resources/generate_data.scala 1000 > target/data
- run the app in local mode
java -cp target/uja-0.1.0-SNAPSHOT-jar-with-dependencies.jar com.twitter.scalding.Tool com.github.sorhus.uja.UserJourneyAnalysis --local --input target/data --output-nodes target/nodes --output-edges target/edges
- combine the output into a dot file and generate the graph
./src/main/resources/generate_graph.sh
- open target/uja.svg in your favorite browser
The end result will look something like this.
