Nova Bank Credit Risk Analytics Dashboard - DataDNA September 2025 Challenge Submission
Project Overview
This submission presents a comprehensive 7-page credit risk analytics platform developed for the DataDNA September 2025 Challenge by Onyx Data. The solution addresses Nova Bank's critical challenge of balancing lending accessibility with risk management across USA, UK, and Canada markets. This platform transforms complex credit data spanning 32,581 clients into actionable insights through innovative visualisation techniques and AI-powered analytics.
Design Philosophy and Approach
Purpose-Driven Architecture
Each page serves a distinct analytical purpose, following a logical workflow from executive overview to granular analysis:
- Page 1 – Overview: Executive decision-making with honeycomb card layout for 30-second portfolio health assessment
- Page 2 – Risk Demographics: Borrower characteristic analysis with cross-dimensional risk matrices
- Page 3 – Financial Risk Analysis: Deep-dive into loan-to-income ratios, debt analysis, and credit patterns
- Page 4 – Market Insights: Geographic risk analysis and loan performance by region and purpose
- Page 5 – Individual Credit Risk: Client-by-client tabular analysis with searchable interface
- Page 6 – Metrics Explorer: Self-service analytics powered by ZoomCharts drill-down capabilities
- Page 7 – AI Assistant: Claude Sonnet 4.0 integration for natural language data exploration
Visual Design Strategy
- Neutral Palette with Strategic Red Highlights: Professional aesthetics ensuring risk alerts command immediate attention while reducing cognitive fatigue during extended analysis sessions.
- Cognitive Design Principles: Leveraging how the brain processes information (motion to colour to text) through pulsing red alerts that capture peripheral vision and create natural priority hierarchies.
Key Innovations and Technical Implementation
1. Executive-Focused Honeycomb Layout
16 metric cards arranged in honeycomb formation deliver pre-processed insights that eliminate analysis paralysis. Each card provides: current state, benchmark comparison, directional indicator, and risk level assessment.
2. Dynamic Risk Matrix System
Cross-dimensional heat maps with red square flash indicators provide real-time pattern recognition across multiple variables:
- Age vs Income risk correlations
- Employment vs Home Ownership matrices
- Education vs Marital Status risk profiles
3. Accessibility-First Tooltips
Comprehensive textual tooltip system explains complex financial metrics in plain language, reducing barriers for non-technical stakeholders and supporting inclusive decision-making.
4. Advanced Tabular Analysis
Individual Credit Risk page features client-by-client breakdown with:
- Custom SVG status tags (Flagged, Clear, Default, Adimplente)
- Searchable and filterable interface
- Colour-coded loan grades and risk indicators
- Revenue impact calculations at individual level
5. Self-Service Analytics Engine
Metrics Explorer powered by ZoomCharts enables:
- Dynamic metric selection across 6 key performance areas
- Cross-dimensional analysis across 9+ segmentation views
- Interactive drill-down capabilities without IT dependency
- Real-time filtering and comparison tools
6. Revolutionary AI Integration
Industry-first Claude Sonnet 4.0 chat assistant embedded directly in Power BI:
- Natural language querying capabilities
- Knowledge base powered by underlying dataset
- Contextual insights and recommendations
- Democratizes data access for non-technical users
Technical Craftsmanship
Hand-Built Power BI Development
Every element constructed line-by-line using advanced techniques:
- DAX: Complex calculations with conditional logic for risk scoring
- HTML/CSS: Custom formatting and responsive design implementation
- SVG: Scalable status indicators and geographic visual elements
- Javascript: Interactive behaviours and animation sequences
- Advanced Power BI: Custom visuals and optimised data modelling
Performance Optimisation
- Efficient data modelling for 32,000+ client records
- Smart aggregations enabling real-time updates
Business Impact and User Experience
For Executives
- 30-second portfolio health assessment through honeycomb overview
- Geographic risk insights enabling regional strategy adjustments
- AI-powered strategic questioning capabilities
For Risk Analysts
- Multi-dimensional risk matrices revealing hidden patterns
- Self-service exploration reducing IT dependency
- Individual client analysis supporting detailed case investigation
For Operations Teams
- Accessible design minimising training requirements
- Familiar tabular interfaces enhanced with visual guidance
- Export capabilities maintaining existing workflow integration
Challenge Requirements Addressed
Core Analysis Questions Answered:
- Borrower default patterns across demographic segments
- Loan purpose risk analysis (education, medical, personal, debt consolidation)
- Loan-to-income and debt-to-income ratio impact on repayment
- Employment type and home ownership risk correlation
- Credit history length effect on loan outcomes
- Geographic risk differences across USA, UK, and Canada
- Loan grade and term safety assessment
- Risk segmentation for "safe" versus "risky" borrower identification
Advanced Insights Delivered:
- Real-time risk threshold monitoring with 40% credit utilisation benchmark
- Interest rate clustering analysis revealing high-risk lending zones
- Payment behaviour correlation with future default probability
- Geographic heat mapping of risk concentration by country/state/city
- Demographic risk profiling across age, education, and employment dimensions
Differentiation and Innovation
This solution represents a paradigm shift in financial analytics, combining:
- Traditional risk management rigour with modern UX principles
- Cutting-edge AI capabilities within established BI frameworks
- Multi-modal interfaces (visual, tabular, conversational) in unified platform
- Accessibility-focused design democratizing complex analytics
- Performance-optimised architecture handling enterprise-scale data
The result is a comprehensive platform serving stakeholders from C-level executives to front-line analysts, maintaining institutional-grade accuracy while revolutionising user experience through thoughtful design and emerging AI technologies.
Technical Specifications
- Platform: Microsoft Power BI with custom visual integration
- Data Volume: 32,581 client records
- Visualisation Library: ZoomCharts for enhanced interactivity + Custom/Native Visuals
- AI Integration: Claude Sonnet 4.0 for natural language processing
- Geographic Coverage: USA, UK, Canada with state/province-level analysis
- Risk Metrics: 16+ key performance indicators with real-time monitoring
eyJrIjoiOTE3MmVjMDAtMjk5Ny00OGZlLTk5OGUtZjhiYmQ1ZWE0M2I5IiwidCI6IjlkOWJiNDE4LTAyY2EtNDFiZC05YTRkLTEwOGE5NmIxM2JmZCJ9
https://purple-forest-02f52071e.1.azurestaticapps.net/iframeEmbed/281648
