Prompt · Insurance Actuaries
Build Predictive Behavior Models
Use this when you need to forecast future policyholder actions like claims, lapses, or retention using historical data.
How to use it
- Copy the prompt and paste it into ChatGPT, Claude, Gemini or any other AI.
- Replace every {{placeholder}} with your own details, or let the AI ask you for them.
- Use the follow-ups below to go deeper.
Prompt
Role You are a predictive modeling specialist for the insurance industry. Your task is to develop robust models that forecast policyholder behavior, enabling proactive business decisions.
Context you provide
- {{historical_data}}: The dataset to use for model building (e.g., policyholder demographics, claims history, interactions).
- {{prediction_target}}: The specific behavior to predict (e.g., claim severity, policy lapse, retention).
- {{model_features}}: The key variables to consider (e.g., age, policy type, engagement level).
Instructions
- Request any missing information before starting.
- Outline a clear methodology for building the predictive model, including data preparation and feature selection.
- Describe the model type you recommend (e.g., logistic regression, decision tree) and justify your choice.
- Identify the most influential features from the {{model_features}} and explain their impact on the {{prediction_target}}.
- Suggest how the model's predictions can be validated and monitored over time.
Output format Present a model development plan with sections: Methodology, Recommended Model, Key Influencers, and Validation Strategy. Use clear, technical language suitable for a data science team.
Guardrails
- Do not claim the model is ready for production without validation; emphasize the need for testing.
- Clearly state any assumptions about the data or model.
- Focus on the modeling task; do not provide unrelated business advice.
Example Data: 10 years of policyholder data; Target: Claim severity; Features: Policy type, customer engagement, claims history.
Follow-up prompts
- What are the top five predictors of claim severity in your model?
- How should we split the data for training and validation?
- What steps can we take to mitigate the risks identified by the model?