Implementation:Evidentlyai Evidently Grafana Dashboard Config
| Knowledge Sources | |
|---|---|
| Domains | ML_Monitoring, Visualization |
| Last Updated | 2026-02-14 12:00 GMT |
Overview
External tool documentation for Grafana dashboard JSON configurations used with Evidently metric storage in PostgreSQL.
Description
Evidently provides example Grafana dashboard JSON configurations that visualize metrics stored in PostgreSQL. The dashboards are provisioned via Grafana's JSON model and include panels for drift scores, missing values, model quality metrics, and LLM evaluation scores.
The dashboard configurations are designed for use with Docker Compose setups that provision PostgreSQL and Grafana together.
Usage
Deploy via Docker Compose with the provided configuration files. Import the dashboard JSON into Grafana or use the provisioning mechanism.
Code Reference
Source Location
- Repository: evidently
- File: examples/data_drift_grafana_dashboard/dashboards/data_drift.json
- File: examples/llm_eval_grafana_dashboard/dashboards/chatbot_evals.json
- File: examples/data_drift_grafana_dashboard/docker-compose.yml
Configuration
# docker-compose.yml structure
services:
db:
image: postgres
environment:
POSTGRES_PASSWORD: example
grafana:
image: grafana/grafana
ports:
- "3000:3000"
volumes:
- ./dashboards:/var/lib/grafana/dashboards
- ./config/grafana_datasources.yaml:/etc/grafana/provisioning/datasources/datasource.yaml
- ./config/grafana_dashboards.yaml:/etc/grafana/provisioning/dashboards/dashboard.yaml
Import
# No Python import - external tool
cd examples/data_drift_grafana_dashboard
docker compose up -d
# Grafana at http://localhost:3000
I/O Contract
Inputs
| Name | Type | Required | Description |
|---|---|---|---|
| PostgreSQL metrics | SQL tables | Yes | Metric values stored via psycopg INSERT |
| Dashboard JSON | JSON file | Yes | Grafana dashboard configuration |
| Docker Compose | YAML | Yes | Infrastructure orchestration |
Outputs
| Name | Type | Description |
|---|---|---|
| Grafana dashboard | Web UI | Interactive monitoring panels at http://localhost:3000 |
Usage Examples
Deploy Drift Monitoring Dashboard
cd examples/data_drift_grafana_dashboard
docker compose up -d
python evidently_metrics_calculation.py
# Open http://localhost:3000 for Grafana dashboard
Deploy LLM Evaluation Dashboard
cd examples/llm_eval_grafana_dashboard
docker compose up -d
python evidently_metrics_calculation.py
# Open http://localhost:3000 for Grafana dashboard