Dashboard is one part of the PEACH project which allows to collect and visualize metrics about recommendation system. It can be generic metrics like number of collected events, or more business oriented to evaluate different recommendation algorithms, for example, click-through rate or diversity of recommended content.
- Pipe task is running in the Spark cluster regularly to calculate the metrics based on collected events. After being calculated metrics are stored in InfluxDB.
- InfluxDB is open source time series database which is used to store the metrics.
- Grafana is open source web application which allows to query and to visualize data from differnet sources including InfluxDB.
There is existing pipe task which provides generic metrics for all organisations participating in the project.
In case custom metrics are required separate task should be created. Check pipe-task documentation for the details.
You can see the example for metric calculation here
It is important to understand key concepts of InfluxDB described here.
Also check the guidelines on schema design
For example, The generic metrics algorithm provides next information:
- Measurement event_types with tags codops, metric and one field - value. Contains metrics about number of different event types collected per organisation.
If you would like to get an account to access https://peach.ebu.io/dashboards, please, contact the PEACH team.