Interactive UI¶

The RWA Calculator includes a web-based interactive interface for running calculations, exploring results, and referencing regulatory parameters—all without writing code.

Prerequisites¶

Before using the UI, ensure you have installed the calculator with UI dependencies:

# Install with UI support
pip install rwa-calc[ui]

# Or with uv
uv add rwa-calc --extra ui

This installs Marimo and Uvicorn, which power the web interface.

Starting the UI Server¶

Launch the multi-application server:

# Using the console command (recommended)
rwa-calc-ui

# Or using the module directly
python -m rwa_calc.ui.marimo.server

# Or using uvicorn
uvicorn rwa_calc.ui.marimo.server:app --host 127.0.0.1 --port 8000

# Using uv
uv run python src/rwa_calc/ui/marimo/server.py

# Or using uvicorn
uv run uvicorn rwa_calc.ui.marimo.server:app --host 127.0.0.1 --port 8000

Once started, open your browser to http://localhost:8000.

Available Applications¶

The UI provides four integrated applications plus an editable workbench:

RWA Calculator¶

The main calculator application at http://localhost:8000/ allows you to run RWA calculations through a visual interface.

Configuration Options¶

Running a Calculation¶

1. Set your data path - Enter the path to your data directory
2. Select format - Choose Parquet or CSV
3. Choose framework - CRR for current rules, Basel 3.1 for future rules
4. Configure options - Set reporting date and IRB toggle
5. Run calculation - Click the calculate button

Understanding Results¶

The calculator displays:

• Summary Statistics

  • Total EAD (Exposure at Default)
  • Total RWA (Risk-Weighted Assets)
  • Average Risk Weight (RWA / EAD)

• Breakdown by Approach

  • Standardised Approach RWA
  • IRB RWA (if enabled)
  • Slotting RWA (for specialised lending)

• Performance Metrics

  • Calculation duration
  • Throughput (exposures per second)

• Results Preview

  • First 100 rows of detailed results
  • Export to CSV option

Results Explorer¶

The Results Explorer at http://localhost:8000/results provides interactive analysis of calculation outputs.

Filtering Options¶

Aggregation Views¶

Aggregate results by different dimensions:

• By Exposure Class - See totals for each exposure category
• By Approach - Compare SA vs IRB vs Slotting
• By Risk Weight Band - Distribution across risk weight ranges

Exporting Data¶

Export your filtered and aggregated results:

• CSV - For spreadsheet analysis
• Parquet - For further processing with Polars/Pandas

Framework Reference¶

The Framework Reference at http://localhost:8000/reference provides an interactive regulatory reference.

Available Sections¶

This reference is useful for:

• Validating calculation inputs
• Understanding regulatory differences
• Quick lookup of risk weights and parameters

Workbench¶

The Workbench at http://localhost:8002/ provides an editable environment for interactive analysis using Python and SQL.

How It Works¶

The template applications (Calculator, Results Explorer, etc.) are read-only — you can view and interact with their outputs but cannot modify code. The Workbench lets you duplicate any template into your own editable workspace.

Duplicating a Template¶

Use the REST API to duplicate a template:

# Duplicate the calculator template
curl -X POST "http://localhost:8000/api/workbooks/duplicate?template=calculator"

# Duplicate with a custom name
curl -X POST "http://localhost:8000/api/workbooks/duplicate?template=results_explorer&name=my_analysis"

The response includes the URL to open the workbook in the Workbench.

Managing Workbooks¶

# List available templates
curl http://localhost:8000/api/templates

# List your workbooks
curl http://localhost:8000/api/workbooks

# Delete a workbook
curl -X DELETE http://localhost:8000/api/workbooks/my_analysis

Using the Workbench¶

Once a workbook is duplicated, open it in the Workbench to:

• Edit Python cells — modify calculations, add visualisations, explore data
• Add SQL cells — query Polars DataFrames with SQL using mo.sql()
• Save changes — workbooks persist in workspaces/local/ across server restarts

Data Requirements¶

The UI expects data in the same format as the Python API. Place your files in a directory structure:

your_data_directory/
├── counterparty/
│   └── counterparties.parquet
├── exposures/
│   ├── facilities.parquet
│   └── loans.parquet
├── collateral/           # Optional
│   └── collateral.parquet
├── guarantee/            # Optional
│   └── guarantee.parquet
└── ratings/              # Optional
    └── ratings.parquet

See Input Schemas for detailed field requirements.

Troubleshooting¶

Server won't start¶

Error: ModuleNotFoundError: No module named 'marimo'

Install with UI dependencies:

pip install rwa-calc[ui]

Error: Port 8000 already in use

Use a different port:

uv run uvicorn rwa_calc.ui.marimo.server:app --port 8080

Data path not found¶

• Ensure the path is absolute or relative to your current working directory
• Check that the required files (counterparties, facilities, loans) exist
• Verify file format matches your selection (Parquet vs CSV)

Calculation errors¶

• Check the error panel for specific validation failures
• Ensure required fields are present in your data
• See Data Validation for field requirements

Next Steps¶

• Configuration Guide - Advanced configuration options
• Calculation Methodology - Understanding how RWA is calculated
• Data Model - Detailed schema documentation