Evidence matrix for soft computing in sequential credit scoring and dynamic credit decision support

Description

This dataset accompanies the review article 'Soft computing for sequential credit scoring and dynamic credit decision support: A critical review'. It contains article-level coding for the 96 credit-specific core sources used in the evidence synthesis. The coding covers citation metadata, primary and alternative method classifications, secondary method tags, decision layer, data structure, task, evidence class, evaluation dimensions, governance relevance, claim-boundary limitations, coding rationale, source location, and audit status. The XLSX workbook contains README, Evidence Matrix, Summary, and Codebook sheets. The CSV file is a machine-readable export of the Evidence Matrix. The files contain no borrower-level or personal data.

Steps to reproduce

Use the Evidence Matrix sheet in the XLSX workbook, or the equivalent CSV file, as the article-level source table. Each of the 96 core sources appears once under its primary method family and evidence class. Reproduce the main cross-tabulation by grouping records by 'Primary method family' and 'Evidence class'. Use 'Alternative classification' and 'Classification sensitivity' to reproduce the method-family sensitivity check. The Summary sheet reports reconciliation totals, and the Codebook defines every field. Evidence classes are non-ordinal and must not be treated as study-quality ranks.

Institutions

• James Cook University

  Queensland, Townsville

Categories

Licence