This project is an end-to-end Model Monitoring Dashboard that tracks performance degradation, detects drift (concept & covariate drift), and allows you to retrain models directly from the UI.
Note: The data used in this project is synthetically generated for demonstration purposes. Therefore, the performance metrics (MSE, R²) may not reflect realistic model performance and should not be interpreted as indicators of actual predictive power. The goal of this project is to demonstrate the monitoring and drift detection workflow, not model accuracy.
We use Streamlit for the frontend and Evidently AI for drift detection. This is a complete monitoring workflow designed for real-world production pipelines.
.
├── dashboard.py # Streamlit Dashboard
├── generate_data.py # Synthetic Data Generator
├── train_model.py # Model Training Script
├── monitor.py # Generates Evidently Report & Feature Importance Drift
├── reference_data.csv # Historical (Training) Data
├── current_data.csv # Current (Production) Data
├── plots/ # Contains images for drift visualizations
│ ├── feature_importance_drift.png
├── demo/ # Contains demo video
│ ├── dashboard-demo.mp4
└── README.md # This file
git clone https://github.com/snehsuresh/ModelMonitoringDashboard.git
cd ModelMonitoringDashboardpython generate_data.pypython train_model.pypython monitor.pystreamlit run dashboard.pyCompare performance (MSE, R²) between reference and current datasets across a selected date range.
Visualize the detailed Evidently AI Drift Report directly within the dashboard.
See real-time performance degradation alerts based on drift.
Compare target vs prediction across time.
See detailed drift metrics directly in the dashboard.
Track how the importance of features changed between reference and current data.
Select any date range to analyze how performance and drift evolve over time.
If drift or degradation is detected, you can trigger retraining directly from the dashboard.
+-------------------+ +-------------------+ +------------------+
| Reference Data | | Current Data | | Model (RandomForest) |
| (Historical) | | (Production) | | Trained on Reference |
+-------------------+ +-------------------+ +------------------+
\ /
\ /
\ /
+---------------------+
| Evidently AI |
| Drift Detection |
+---------------------+
|
v
+---------------------+
| Streamlit UI |
| Real-Time Monitoring |
+---------------------+
|
v
+---------------------+
| Retrain Trigger |
+---------------------+
MIT License