Algorithmic Workflow: Bank Marketing Customer Conversion Forecasts

Fahim Nasir published a dataset on figshare on May 15, 2026, detailing a workflow for forecasting customer conversion in bank marketing. The dataset is 17.5 KB in size and is stored in an XLS file format. It was created to support a study proposing a two-stage quadrilateral evaluation framework for model selection under class imbalance.

Use Cases

• Evaluating sampling techniques like Borderline2SMOTE based on their impact on model calibration and fairness.
• Comparing ensemble and deep learning models based on discrimination, computational cost, and explainability metrics.
• Applying SHAP (Shapley Additive exPlanations) for stable interpretability to identify key drivers of customer responses.
• Selecting suitable model-sampling combinations for responsible predictive analytics in Banking 4.0.

Strengths

• Dataset is openly licensed under CC-BY-4.0.
• Workflow includes a specific evaluation framework assessing discrimination, calibration, cost, and explainability.
• Explicitly tests XGBoost with Borderline2SMOTE sampling.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Dataset is very small at 17.5 KB, indicating limited scope.

Provenance

Source

figshare

Collection Method

Likely contains results from a two-stage evaluation of models with sampling techniques.

Freshness

Last updated 2026-05-15 17:44:00; freshness should be verified.