METARFM: A META-LEARNING FRAMEWORK FOR THE ADAPTIVE SELECTION OF RFM MODEL ARIANTS IN CUSTOMER SEGMENTATION
Bibliographic record
Abstract
The Recency-Frequency-Monetary (RFM) model is a widely used method for customer segmentation, but its effectiveness depends on selecting the appropriate variant (e.g., weighted or entropy-based) for a given dataset. This selection process is typically manual and task-specific, leading to inconsistent results and limited generalizability. To address this issue, we present MetaRFM, a novel automated framework for selecting optimal RFM variants. MetaRFM mines a set of meta-features—such as sparsity, diversity, and skewness—extracted from customer transaction datasets, including both personal transaction data and product purchase information. These meta-features characterize the dataset at a high level, enabling the framework to predict which RFM variant would perform best. A meta-learner is trained to map these meta-features to the performance of different RFM variants, which are evaluated using both cluster quality metrics (Silhouette Score, Davies-Bouldin Index) and business-relevant metrics (predictive lift, churn prediction accuracy). Extensive experiments conducted on real-world datasets from retail, e-commerce, and subscription services show that MetaRFM consistently outperforms static and single-variant models. On average, MetaRFM improves cluster separation by 15.7% and campaign lift by 22.3%. This framework provides a systematic, scalable solution for selecting the most appropriate RFM model, improving segmentation robustness and business relevance. The results highlight the substantial potential of meta-learning for adaptive, context-aware analytics in marketing, offering a more effective approach to customer segmentation and optimizing marketing strategies.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.007 | 0.014 |
| Meta-epidemiology (narrow) | 0.003 | 0.001 |
| Meta-epidemiology (broad) | 0.003 | 0.004 |
| Bibliometrics | 0.004 | 0.002 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.003 |
| Open science | 0.005 | 0.002 |
| Research integrity | 0.003 | 0.004 |
| Insufficient payload (model declined to judge) | 0.003 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".