A Hybrid Data-Driven and Fuzzy MCDM Approach for Employee Selection

Employee selection, a cornerstone of human resource management, critically shapes organizational performance and long-term effectiveness. While traditional approaches primarily rely on expert-based evaluations, this study proposes a novel hybrid framework that integrates Multi-Criteria Decision-Making methods with data mining techniques to reduce the dimensionality of the number of criteria or variables considered. By integrating backward regression with fuzzy Multi-Criteria Decision-Making methods, our framework reduces model complexity and captures criteria interdependencies, while fuzzy logic addresses ambiguity in expert judgment, a gap often overlooked in prior research. The methodology first uses backward regression modeling with the employee attrition rate as the response variable to identify core criteria. Subsequently, the fuzzy Decision-Making Trial and Evaluation Laboratory analyzes interrelationships between criteria, followed by the fuzzy Analytic Network Process for weighting criteria and ranking candidates. We validate our approach using real-world recruitment data—including expert interview scores and historical attrition—from a company specializing in electronic attendance systems. The AI-generated rankings are benchmarked against these expert-based evaluations to assess alignment with human judgment. Initially, 17 criteria were systematically reduced to 11 core factors, resulting in a streamlined yet robust evaluation system. Our findings emphasize that “Time-of-service,” “Requested-wage,” “Teamwork,” and “Leadership” are the most critical criteria influencing effective IT personnel selection.