This professional-grade prompt is designed to elicit a modular, production-ready code structure from an AI model. You can copy and paste the text below directly into your prompt window.

Prompt: Professional Data Cleaning Pipeline Architect

🎭 Role

Act as a Senior Data Engineer and Python Architect specializing in scalable data pipelines. Your code must prioritize performance, maintainability, and clean code principles (PEP 8). You are an expert in pandas, numpy, and defensive programming techniques.

🌐 Context

We are working with a [SCENARIO, e.g., raw financial transaction dataset] that contains significant noise, inconsistent formatting, and missing values. The goal is to transform this messy CSV into a "gold-standard" dataframe ready for machine learning or analytical reporting. The pipeline must be highly readable and follow functional programming paradigms, specifically utilizing pandas method chaining.

🛠️ Task Instruction

Construct a robust, modular Python class or function-based pipeline that executes the following steps in sequence:

1. Duplicate Management: Identify and drop duplicate rows based on a user-defined list of [COMPOSITE_KEYS].
2. Imputation Logic: Implement a flexible strategy for missing values:
   • Forward-fill and backward-fill for time-series columns.
   • Mean/Median imputation for numerical features.
3. Outlier Mitigation: Utilize the Interquartile Range (IQR) method to identify and cap/remove outliers per column.
4. Temporal Standardization: Normalize all date-time columns into a standard [DATE_FORMAT] format.
5. Quality Assurance: Generate a summary report (pre- and post-processing) that includes:
   • Total rows dropped.
   • Percentage of missing values handled.
   • Summary of outlier counts per feature.

⚖️ Constraints & Tone

• Tone: Professional, technical, and instructive.
• Best Practices:
  • Use .pipe() or explicit method chaining to ensure readability.
  • Avoid modifying the original dataframe (return copies to prevent SettingWithCopyWarning).
  • Include docstrings for all functions/methods.
  • Include error handling (e.g., check if columns exist before processing).
• Avoid: Hard-coding variable names where possible; favor configuration dictionaries or arguments.

📝 Output Format

1. Pipeline Implementation: A clean, modular code block containing the processing functions.
2. Usage Example: A brief snippet demonstrating how to initialize the pipeline with a sample CSV path.
3. Data Quality Logic: A clearly defined function that prints the "Before/After" report.
4. Explanation: A short, bulleted section explaining why specific pandas methods were chosen for performance.

Configuration Variables

• Input CSV Path: [PATH_TO_CSV]
• Target Columns: [LIST_OF_COLUMNS_TO_PROCESS]
• Composite Keys: [LIST_OF_KEYS_FOR_DUPLICATION]
• Date Columns: [LIST_OF_DATE_COLUMNS]

How to use this prompt:

2. Replace the bracketed variables (e.g., [SCENARIO], [PATH_TO_CSV]) with your specific project details.
3. Paste the completed text into your AI interface.