How subham-debnath/financial-fraud-detection works
This is a Streamlit credit-card fraud detection application. Users can review a dashboard, upload transaction data for batch predictions, score a single transaction, inspect prediction history, and view model information. The app also supports model selection, probability thresholds, visual summaries, and CSV downloads. The sampled source is a single large application file, so the graph separates its user-facing responsibilities without implying separate implementation modules. Some model-loading and prediction internals fall in omitted excerpts; those links are marked unverified.
Main entry points: app.py Page navigation · models Serialized models · model.pkl Bundled model
Components 12
Application interface
- Streamlit app
app/app.py - Page navigation
app/app.py
Fraud scoring
- Batch upload
app/app.py - Single transaction
app/app.py - Model selection
app/app.py - Serialized models
models - Bundled model
app/model.pkl
Results and insights
- Fraud dashboard
app/app.py - Prediction history
app/app.py - Model information
app/app.py - CSV exports
app/app.py
Other components
- User
Connections 12
- User → Streamlit app: uses
app/app.py - Streamlit app → Page navigation: offers pages
app/app.py - Page navigation → Batch upload: opens
app/app.py - Page navigation → Single transaction: opens
app/app.py - Page navigation → Fraud dashboard: opens
app/app.py - Page navigation → Prediction history: opens
app/app.py - Page navigation → Model information: opens
app/app.py - Streamlit app → Model selection: selects model
app/app.py - Model selection → Serialized models: loads from
- Model selection → Bundled model: may load
- Batch upload → CSV exports: downloads results
app/app.py - Batch upload → Prediction history: records summary
app/app.py