Optimizing HR Decision-Making for Recruitment and Employee Performance with Hybrid Deep Learning Models

In HR, algorithmic decision-making is rapidly emerging as a novel tool for development and recruitment. As digitisation accelerates in enterprises, the utilisation of algorithms in HRM is increasing and will assume an even more pivotal position in the future. Both overt and covert prejudice provide a risk of unjust treatment towards particular demographic groups, notwithstanding their potential to assist firms in minimising expenses and enhancing objectivity. The suggested methodology comprises three stages: data preparation, feature selection, and model training. Before feature selection, data is refined by eliminating noise, outliers, and missing values with the Best-Worst Method. The study introduces two novel methodologies for HRM forecasting, applicable to the recruitment process, utilising a hybrid ARIMAXG model. The efficacy of ARIMAXG in human resource decision-making for recruitment was evidenced by its optimal accuracy rate of 96.56%, attained by one of the four evaluated models. This finding indicates that the strategy can substantially improve the predictive accuracy of HR forecasting. The human resources development and recruitment procedures depend significantly on algorithmic decision-making because of its numerous benefits, such as enhanced impartiality and cost efficiency. Organisations must consistently monitor for indicators of bias and discrimination. The enhanced accuracy of the ARIMAXG model underscores its potential as a reliable instrument for optimising HR decision-making in the recruitment process.