How soumajit11/cosmic-classifier works
Cosmic Classifier is a Streamlit planetary-habitability explorer: users configure ten telemetry values, receive a model classification, and inspect charts or export the inputs. The dashboard entry point loads the serialized classifier and routes landing, input, and report views. The README also documents a notebook-based training and model-export workflow. Training internals and some report-view code are not sampled; those documented capabilities are represented at the notebook level, without assuming unseen implementation details.
Main entry points: views.py Page views · app.py App router · decision_tree_model.pkl Habitability classifier · model.ipynb Training workflow
Components 12
Model training
- Training workflow
model.ipynb - Planetary datasets
Dashboard flow
- App router
app.py - Page views
src/views.py - Model utilities
src/model_utils.py - Habitability classifier
decision_tree_model.pkl - Live inference
src/views.py - CSV export
src/views.py
Reports and presentation
- Theme styling
src/styles.py - Classification details
src/model_utils.py - Telemetry charts
src/charts.py
Other components
- Explorer
Connections 13
- Explorer → App router: uses dashboard
- App router → Theme styling: loads styling
- App router → Model utilities: loads model
- Model utilities → Habitability classifier: loads pickle
- App router → Page views: routes pages
- Page views → Live inference: predicts
- Live inference → Habitability classifier: runs model
- Page views → Classification details: renders result
- Page views → Telemetry charts: builds charts
- Page views → CSV export: creates download
- Explorer → CSV export: downloads CSV
- Planetary datasets → Training workflow: supplies data
- Training workflow → Habitability classifier: trains and saves