Detect data drift and real time monitoring loss function
Tools:
Evidently
- library for the metrics calculationsPrometheus
- service to store the metricsGrafana
- service to build the dashboardsDocker
- service to rule them all
- Install Docker if you haven't used it before.
- Build container using
docker compose up
- Install dependencies.
- install dependencies
pip install -r requirements.txt
- Then run the docker image from the example directory:
./scripts/example_run_request.py
- Install Docker if you haven't used it before.
- Build container using
docker compose up
- Prepare your data as dict format and send it via endpoint(http://localhost:8085/iterate/data-car)
- Explore the dashboard.
Go to the browser and access the Grafana dashboard at http://localhost:3000. At first, you will be asked for login and password. Both are admin
.
To see the monitoring dashboard in the Grafana interface, click "General" and navigate to the chosen dashboard (e.g. "Evidently Data Drift").
- Stop the example.
To stop the process of sending data, cancel the execution of the example script (press CTRL-C).
If the Docker containers were not stopped after that, run a command:
docker compose down