Skip to main content
Back to challenges
Challenge

Multifamily Property Risk Investment Premium Reduction Scoring Model

Multifamily property owners invest heavily in safety upgrades (e.g. water sensors, security cameras) that often fail to reflect in insurance premium discounts. Build a predictive loss reduction model using LightGBM and spatial building data to quantify premium reduction credits accurately.

Machine LearningHosted by Vera
Sign in free to start
Challenge brief

What you are building

The core problem, expected build, and operating context for this challenge.

Develop a LightGBM loss reduction scoring engine to translate property safety investments into verifiable premium discounts.

Delivery guide

How work is evaluated

Evaluation

Evaluates predicted claim frequency reduction precision and discount rate assignment.

Datasets

Challenge datasets

Public datasets, plus the private datasets you have access to.

Evaluation rubric

How submissions are scored

These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.

Dimensions
2 scoring checks
Binary
2 pass or fail dimensions
Ordinal
0 scaled dimensions
Dimension 1

savings_math

Validates annual_savings_usd equals base_annual_premium times recommended_discount_pct

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Dimension 2

discount_gini_score

Gini score of loss frequency prediction • target: 0.45 • range: 0-1

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Learning goals

What you should walk away with

  • Feature engineer IoT sensor installation data (water shutoff valves, smart smoke detectors)

  • Train LightGBM loss frequency models on multifamily property loss histories

  • Calculate actuarially sound premium discount percentages per safety technology installed

  • Validate model calibration across regional property hazard tiers

Resources and assets

Reference links and supporting material

Dataset notes

Multifamily property characteristics dataset with IoT safety technology metadata and historical loss records.

Challenge setup

Evaluates predicted claim frequency reduction precision and discount rate assignment.

Challenge input

JSON containing property characteristics and active safety investments.

LightGBM

Gradient boosting framework for tabular loss data

Evaluated output

JSON containing expected_loss_reduction_pct, recommended_discount_pct, and roi_years.

Evaluation checks
  • Validates annual_savings_usd equals base_annual_premium times recommended_discount_pct
  • Gini score of loss frequency prediction • target: 0.45 • range: 0-1
Configured evaluation evidence
  • Benchmark: MultifamilyPricingBench
  • Discount Gini Score target: 0.45
  • 1 public reference case
Configured runtime
  • Python execution harness
  • Python sandbox (unavailable on Versalist)

Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.

View technical recipe

Configured tools

Action Space
  • Anthropic Claude · Required
  • Mistral · Optional
Policy Serving
  • Groq Cloud · Optional

Evaluation contract

  • savings_math · Weight 1
  • discount_gini_score · Weight 1

Metrics

  • discount_gini_score · Target 0.45

Benchmarks

  • MultifamilyPricingBench
  • general

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON containing property characteristics and active safety investments.

    Output: JSON containing expected_loss_reduction_pct, recommended_discount_pct, and roi_years.

    Execution harness: Python

Evaluation tools

  • LightGBM · ML Framework

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Multifamily Property Risk Investment Premium Reduction Scoring Model