How ditheshshettigar00-jpg/geopovertyai works
GeoPoverty AI estimates district-level deprivation from satellite-derived indicators, compares estimates with NFHS-5 survey benchmarks, and exposes results through a web dashboard. The documented workflow runs from geospatial feature extraction through leakage-aware state-held-out modeling to district predictions and decision-support views. The graph also includes authentication, reporting, and indicator services. Some links reflect README-documented capabilities rather than inspected call sites; unsampled modules are represented only where the README or source index supports their role.
Main entry points: routes.py Prediction API · app.py Flask application · reports.py District reports · model_service.py Model service
Components 23
Geospatial data
- Satellite features
geospatial/extract_geospatial_features.py - District boundaries
geospatial/fetch_district_boundaries.py - Name matching
preprocessing/name_keys.py
Modeling and explanation
- Feature pipeline
training/pipelines.py - Spatial validation
training/geo_validation.py - Model training
training/train_models.py - SHAP explanations
explainability/shap_explain.py
Application services
- Flask application
backend/app.py - Prediction API
backend/api/routes.py - Model service
backend/services/model_service.py - District indicators
backend/services/indicator_service.py - District profiles
backend/services/profile_service.py - Hotspot clustering
backend/services/hotspot_service.py - Hidden gap detector
backend/services/residual_service.py - District reports
backend/api/reports.py - Authentication
backend/api/auth.py - Administration
backend/api/admin.py - SQLite data
backend/services/db_service.py
Dashboard
- Choropleth map
frontend/static/js/map.js - Scenario explorer
frontend/static/js/predict.js - District comparison
frontend/static/js/compare.js - Hotspot view
frontend/static/js/hotspots.js
Other components
- Researcher
Connections 30
- Researcher → Flask application: uses dashboard
- District boundaries → Satellite features: supplies districts
- Satellite features → Feature pipeline: provides indicators
- Name matching → District indicators: matches district names
backend/services/indicator_service.py - Feature pipeline → Model training: preprocesses features
training/pipelines.py - Spatial validation → Model training: validates state-held-out
- Model training → Model service: supplies fitted model
- Model service → SHAP explanations: explains estimates
backend/services/model_service.py - Flask application → Prediction API: registers API
backend/app.py - Flask application → Authentication: registers authentication
backend/app.py - Flask application → District reports: registers reports
backend/app.py - Flask application → Administration: registers administration
backend/app.py - Prediction API → Model service: requests predictions
backend/api/routes.py - Prediction API → Hotspot clustering: requests clusters
backend/api/routes.py - Prediction API → Hidden gap detector: requests residual atlas
backend/api/routes.py - Prediction API → District profiles: requests profiles
backend/api/routes.py - Prediction API → District indicators: requests indicators
backend/api/routes.py - Prediction API → SQLite data: reads area data
backend/api/routes.py - District reports → SQLite data: reads report data
backend/api/reports.py - District reports → Model service: uses estimates
backend/api/reports.py - District reports → District indicators: uses indicators
backend/api/reports.py - Authentication → SQLite data: reads and writes accounts
backend/api/auth.py - Administration → SQLite data: manages accounts
- Choropleth map → Prediction API: fetches map data
- Scenario explorer → Prediction API: requests scenarios
- District comparison → Prediction API: requests district data
- Hotspot view → Prediction API: requests hotspots
- Researcher → District reports: requests district reports
- Researcher → Hidden gap detector: reviews gap candidates
- Researcher → District indicators: reviews district indicators