An adaptive neuro-fuzzy decision support method for anticipating turnover intention in information technology employees

Employee turnover in the information technology (IT) sector poses substantial organizational challenges, particularly in knowledge-intensive environments characterized by high labor mobility. This study develops and empirically evaluates a hybrid predictive framework that integrates Ant Colony Optimization (ACO) for discrete feature subset selection with an Adaptive Neuro-Fuzzy Inference System (ANFIS) for nonlinear modeling of turnover intention. Using survey data collected from seventy-two IT professionals employed across fifteen Iranian companies, nineteen theoretically grounded predictors were initially identified based on prior literature and expert consultation. The ACO procedure reduced the feature space to five variables that minimized cross-validated prediction error: job satisfaction, organizational commitment, alternative job opportunities, compensation, and job security. The resulting ANFIS model achieved a mean absolute error of approximately 0.30 and a coefficient of determination of 0.88 on a held-out test set, indicating substantial explanatory power for a subjective, continuous construct measured on a Likert-type scale ranging from one to five. Beyond predictive accuracy, the neuro-fuzzy structure enabled extraction of interpretable linguistic rules and response surfaces, allowing examination of nonlinear and interaction effects among key predictors. In particular, joint effects of job satisfaction and perceived alternative opportunities were found to substantially shift predicted turnover intention levels. The findings contribute a sector-specific, interpretable modeling framework for turnover intention assessment within the Iranian IT context. However, validation is limited to a single contextual dataset; further evaluation on standardized public HR benchmarks and multi-organizational samples would be required to assess cross-context generalizability. • Identify key organizational and labor market factors that drive turnover intention among IT employees. • Develop an analytical decision-making method that accurately predicts turnover intention among the IT workforce. • Apply a neuro-fuzzy modeling approach that offers both predictive accuracy and interpretability. • Reveal influential predictors through feature selection to support evidence-based retention decisions. • Generate actionable decision insights that help guide targeted and effective retention strategies.