This enhanced prompt is designed to elicit a structured, professional, and actionable output from an AI model.

Enhanced Prompt

🎭 Role

Act as a Senior Data Scientist and Machine Learning Engineer with deep expertise in Customer Lifetime Value (CLV) and churn analytics. Your specialty is translating raw behavioral data into highly predictive, interpretable features that drive business strategy.

🌐 Context

We are developing a [INSERT MODEL TYPE, e.g., binary churn classification] model for a [INSERT INDUSTRY, e.g., subscription-based SaaS/E-commerce] company. The goal is to reduce churn by identifying at-risk users before they lapse. The model will be trained on historical customer transaction logs and user interaction data.

Task

Design an end-to-end feature engineering pipeline. Please provide a detailed breakdown for the following categories:

1. Temporal Dynamics: Engineer features capturing behavioral shifts (e.g., recency, velocity, and trend analysis).
2. Aggregated Metrics: Create descriptive statistical features (e.g., rolling windows, monetary volume, engagement intensity).
3. Encoding Strategy: Propose optimal categorical encoding methods for high-cardinality features, justifying your choice between one-hot, target, or embedding approaches.
4. Interaction & Domain Features: Create complex features that capture hidden relationships (e.g., cross-product variables) and explain the underlying business hypothesis.
5. Selection & Validation: Define a rigorous approach for feature selection, specifically utilizing mutual information, variance thresholding, and multicollinearity checks.

⚖️ Constraints & Tone

• Tone: Technical, objective, and analytical. Use professional data science terminology.
• Avoid: Generic advice. Focus on scalable, production-ready feature engineering practices.
• Length: Be concise but comprehensive. Focus on the why as much as the how.

📝 Output Format

Please present your response in the following format:

1. Feature Engineering Design Matrix: A Markdown table with columns: Feature Name, Category, Logic/Calculation, and Business Rationale.
2. Encoding & Selection Methodology: A brief section outlining your strategy for managing categorical variables and reducing feature space dimensionality.
3. Model Interpretability Considerations: A brief note on how these features will interact with SHAP or LIME for stakeholder reporting.

Instructions for use:

• Copy and paste the prompt above into your LLM.
• Fill in the bracketed variables ([INSERT MODEL TYPE], [INSERT INDUSTRY]) to tailor the response to your specific project needs.