[heatmap-loss-triangle] Actuarial Loss Development Triangle

[MarkusNeusinger]

Description

A triangular matrix visualization showing cumulative insurance claim payments developing over time. Rows represent accident/origin years, columns represent development periods, with projected (IBNR) values in the lower-right triangle.

Key Visual Elements

• Accident/origin year on rows
• Development period on columns
• Upper triangle: actual cumulative claims (observed)
• Lower triangle: projected/estimated claims (different styling)
• Diagonal: latest evaluation date
• Development factors annotated
• Color coding for actual vs projected values

Applications

• Insurance reserving: estimating IBNR (Incurred But Not Reported) reserves
• Actuarial science: chain-ladder method visualization
• Reinsurance: treaty pricing and reserve adequacy
• Financial reporting: loss reserve disclosure

Domain

Insurance / Actuarial Science

[claude]

✅ Spec ID: heatmap-loss-triangle

Rationale: This is a triangular matrix (heatmap) visualization specific to actuarial loss development, showing cumulative claims across accident years and development periods with distinct styling for actual vs projected values. No existing spec covers this domain-specific visualization.

Add label approved to begin generation.

[github-actions]

✅ Specification Ready: heatmap-loss-triangle

PR: #4602

---

specification.md

heatmap-loss-triangle: Actuarial Loss Development Triangle

Description

A triangular matrix visualization showing cumulative insurance claim payments developing over time. Rows represent accident/origin years, columns represent development periods (e.g., 1-10 years), with the upper-left triangle displaying actual observed cumulative claims and the lower-right triangle showing projected/estimated values (IBNR). This plot is essential for actuarial reserving, enabling analysts to visualize the chain-ladder method and identify development patterns in loss data.

Applications

• Insurance reserving: visualizing cumulative claims development to estimate IBNR (Incurred But Not Reported) reserves using the chain-ladder method
• Reinsurance pricing: assessing reserve adequacy and treaty pricing by examining historical loss development patterns
• Financial reporting: preparing loss reserve disclosures and communicating reserve estimates to stakeholders

Data

• accident_year (integer) - The origin/accident year when claims occurred (e.g., 2015-2024)
• development_period (integer) - The development period in years from origin (e.g., 1-10)
• cumulative_amount (float) - Cumulative claim payments at each development stage
• is_projected (boolean) - Whether the value is actual (observed) or projected (estimated)
• development_factor (float) - Age-to-age development factors between consecutive periods
• Size: 10 accident years x 10 development periods (triangular: ~55 actual + ~45 projected cells)
• Example: Cumulative paid claims triangle with accident years 2015-2024 and development periods 1-10

Notes

• The upper-left triangle contains actual/observed values; the lower-right triangle contains projected values
• Use distinct visual styling (e.g., different background colors or hatching) to clearly differentiate actual from projected cells
• The diagonal from top-right to bottom-left represents the latest evaluation date
• Annotate each cell with the cumulative amount value (formatted with thousands separator)
• Display age-to-age development factors between column headers or as a separate row below the triangle
• Color intensity should encode the magnitude of cumulative amounts
• Include a clear legend distinguishing actual vs projected regions
• Row labels show accident/origin years; column headers show development periods

---

Next: Add approved label to merge this specification to main.

---

🤖 spec-create workflow

[github-actions]

🚀 Specification Merged!

Specification: heatmap-loss-triangle
PR: #4602 (merged)

The specification is now in main and synced to PostgreSQL.

---

Next: Add generate:{library} labels to start generating implementations.

Available libraries: matplotlib, seaborn, plotly, bokeh, altair, plotnine, pygal, highcharts, letsplot

Or use workflow dispatch for bulk operations.

---

🤖 spec-create workflow