Enhancing author assessment: an advanced modified recursive elimination technique (MRET) for ranking key parameters and conducting statistical analysis of top-ranked parameter

Assessing the impact of authors in scientific research is crucial for evaluating scholarly contributions. Various parameters exist in the literature to quantify researchers’ productivity, such as publication count, citation count, and the h index. However, prioritizing the most effective metrics among the plethora available is essential. In this paper, we employ a powerful deep learning technique, the multi-layer perceptron (MLP), for classification and ranking. We propose the MRET technique to assign importance scores to each parameter, aiding MLP in its task. Through comprehensive statistical analysis, we evaluate the top 10 parameters out of 64 distinct metrics using seven well-known statistical methods. Our study utilizes a dataset from the Civil Engineering domain, comprising 590 non-awardees and 590 awardees over the last three decades. Results reveal the normalized h index as the most important parameter among the 64, and the Trigonometric Mean as the superior statistical model. Furthermore, combining parameters such as M-Quotient and FG index with others consistently yields promising results across various statistical models.