How lilibethlangat/opportunity_scout_ai works
Opportunity Scout AI is a startup opportunity decision-support app: users submit a venture profile in the dashboard, the API asks the prediction pipeline to gather market context and score it, then returns explanations and comparisons. The runtime combines aggregation, feature engineering, a gradient-boosting model, and SHAP-based explanations; the UI also supports what-if analysis and portfolio views. The README documents additional live signals and dashboard capabilities, but sampled excerpts do not establish every collector or persistence connection, so uncertain wiring is omitted.
Main entry points: pipeline.py Signal aggregation · main.py FastAPI application · predictor.py Opportunity predictor · script.js Interactive dashboard
Components 21
Dashboard
- Interactive dashboard
Frontend/script.js - Dashboard assets
Frontend/index.html
Application API
- FastAPI application
main.py - Industry / country reference
Data/startup_core.csv - Scored venture data
Data_pipeline/historical_scored.csv
Scoring pipeline
- Opportunity predictor
predictor.py - Signal aggregation
pipeline.py - Feature engineering
feature_engineer.py - LightGBM model
Models/ventures_lightgbm.joblib - SHAP explanations
predictor.py
Market data
- Historical venture data
Data/ventures.csv - Trend history
Data/industry_trends.csv - Macro indicators
Data/macro_data.csv - News signals
Data/GDELT.csv - Reddit signals
Data/reddit.csv
Other components
- Investor / founder
- Google Trends
- Reddit data service
- GDELT / BigQuery
- World Bank / GDP context
- Portfolio reviewer
Connections 25
- Investor / founder → Interactive dashboard: enters venture
- Interactive dashboard → FastAPI application: requests evaluation
- FastAPI application → Opportunity predictor: predicts / compares
- FastAPI application → Industry / country reference: reads options
- FastAPI application → Scored venture data: reads scores
- Opportunity predictor → Signal aggregation: collects signals
- Signal aggregation → Historical venture data: looks up ventures
- Signal aggregation → Trend history: reads trends
- Signal aggregation → Macro indicators: reads macro data
- Signal aggregation → News signals: reads news signals
- Signal aggregation → Reddit signals: reads social signals
- Signal aggregation → Google Trends: fetches trends
- Signal aggregation → Reddit data service: fetches mentions
- Signal aggregation → GDELT / BigQuery: queries news data
- Signal aggregation → World Bank / GDP context: gets GDP context
- Opportunity predictor → Feature engineering: transforms features
- Feature engineering → LightGBM model: prepares model input
- Opportunity predictor → LightGBM model: scores features
- Opportunity predictor → SHAP explanations: explains prediction
- Opportunity predictor → FastAPI application: returns result
- FastAPI application → Interactive dashboard: serves results
- Portfolio reviewer → Interactive dashboard: reviews portfolio
- Interactive dashboard → FastAPI application: requests what-if / compare
- FastAPI application → Scored venture data: returns portfolio view
- FastAPI application → Dashboard assets: serves interface