Skip to content

Releases: SKR-35/Procurement-Risk-Analytics-Shiny

v1.0.0 — Procurement Risk Analytics Dashboard

Choose a tag to compare

@SKR-35 SKR-35 released this 07 Jun 19:51

An end-to-end procurement risk analytics platform built in R/Shiny using public Polish procurement data from Atlas Przetargów.

This release introduces a complete analytical workflow covering procurement risk screening, anomaly detection, relationship analytics, concentration monitoring and interactive dashboard reporting.

Live Dashboard

https://skr-35.shinyapps.io/procurement-risk-analytics-dashboard/

Highlights

Procurement Review Universe

The original Atlas Przetargów dataset contains more than 1.4 million procurement notices covering the full procurement notice lifecycle.

For risk analytics purposes, the project focuses on result and award notices representing concluded procurement outcomes.

This produces:

  • 367,287 result notices
  • 188,788 buyer-vendor relationships
  • 91,730 vendors
  • 17,212 buyers

This filtered review universe provides a more suitable population for procurement risk analysis.

Benford Screening

The platform includes portfolio-level Benford Law screening with:

  • First-digit distribution analysis
  • MAD (Mean Absolute Deviation)
  • Chi-square statistics
  • Province-level segmentation
  • Buyer-level and CPV-level review capabilities

The overall MAD result (~0.0075) indicates acceptable conformity rather than a severe portfolio-wide anomaly.

Vendor Risk Scoring

Vendor risk scores combine indicators such as:

  • Single-offer exposure
  • Low-competition exposure
  • Round-number contract values
  • High-value awards
  • Buyer concentration
  • Relationship concentration

The objective is to prioritize review populations rather than determine wrongdoing.

Buyer Risk Scoring

Buyer scoring evaluates:

  • Vendor concentration
  • Repeat vendor relationships
  • Competition indicators
  • Procurement volume patterns
  • High-value procurement exposure

Minimum-contract filtering is used to reduce instability caused by very small samples.

Relationship Analytics

Buyer-vendor network analysis identifies:

  • Repeated relationships
  • High-value relationships
  • High-frequency relationships
  • Relationship concentration patterns

The project emphasizes that procurement risk is often relationship-based rather than entity-based.

Concentration Monitoring

Concentration analysis evaluates:

  • Buyer dependence on vendors
  • Vendor dependence on buyers
  • Contract concentration
  • Value concentration
  • Dominant relationship patterns

These indicators support procurement review and audit planning.

Regional Risk Monitoring

The dashboard includes voivodeship-level monitoring using:

  • Risk density score
  • Low-competition rate
  • Single-offer rate
  • High-value rate
  • Round-number rate

Interactive geographic visualizations allow regional comparison across Poland.

Dashboard Features

  • Interactive Shiny dashboard
  • Plotly visualizations
  • PDF report generation
  • CSV exports
  • Geographic risk maps
  • Risk scorecards
  • Relationship analytics
  • Concentration monitoring
  • Multiple visual themes

Deployment

The project supports:

  • Local R/Shiny execution
  • shinyapps.io deployment
  • Dockerized deployment

Repository Contents

  • Full analytical pipeline
  • Dashboard application
  • Deployment scripts
  • Docker configuration
  • Step-by-step walkthrough documentation
  • PDF reporting functionality

Important Disclaimer

This project is intended for educational, analytical and risk-screening purposes.

Risk scores, anomaly indicators, concentration metrics and Benford deviations do not constitute evidence of misconduct.

All outputs should be treated as review signals requiring further investigation and professional judgment.

Acknowledgements

Data source:
Atlas Przetargów

Geographic boundaries:
andilabs/polska-wojewodztwa-geojson

Built with:
R, Shiny, Plotly, DT, sf and Docker.