Add an RFM segmentation

[mvanwyk]

Implement Ibis query to calculate customer RFM (Recency, Frequency, Monetary) segmentation scores using a simple method.

SQL Code to implement in Ibis

WITH customer_metrics AS (
  SELECT
    customer_id,
    DATEDIFF(DAY, MAX(transaction_date), CURRENT_DATE) AS recency_days,
    COUNT(DISTINCT transaction_id) AS frequency,
    SUM(unit_spend) AS monetary
  FROM my_table
  GROUP BY customer_id
),

rfm_scores AS (
  SELECT
    customer_id,
    NTILE(10) OVER (ORDER BY recency_days DESC) AS r_score,
    NTILE(10) OVER (ORDER BY frequency DESC) AS f_score,
    NTILE(10) OVER (ORDER BY monetary DESC) AS m_score
  FROM customer_metrics
)

SELECT
  customer_id,
  r_score,
  f_score,
  m_score,
  (r_score * 100) + (f_score * 10) + m_score AS rfm_segment
FROM rfm_scores
ORDER BY rfm_segment DESC;

• The function should allow the user to specify the "current date". ie pass a python date to the Ibis, rather than using the CURRENT_DATE provide by the database.
• Use the column helper (or get_option) for column naming
• Takes in a pandas dataframe or Ibis table
• Add unit tests
• Place the code in segmentation.py for now
• Add an example in the analysis_modules.md like the HML Segmentation
• Use the following description in the function docstring

    Calculates customer RFM segments based on transaction history.
    
    Segments customers using the RFM (Recency, Frequency, Monetary) methodology.
    Each dimension is scored 1-10 (10=best) and combined into a 3-digit segment.
    Higher segment values indicate more valuable customers.
    
    Args:
        { Replace with whatever args you add }
        
    Returns:
        DataFrame with columns:
            - customer_id: Unique customer identifier
            - r_score: Recency score (1-10)
            - f_score: Frequency score (1-10)
            - m_score: Monetary score (1-10)
            - rfm_segment: Combined segment score (range: 111-999)