Abstract:
The employability of graduates has become a measure, for institutions because of the increasing number of graduates
entering the job market and the intense competition for good job opportunities. Many studies have attempted to predict student’s
employability before they graduate using intelligence methods. However implementing these methods has proven to be time consuming
and challenging requires effort with results so far. To address these challenges we propose a technique to identify the factors that
impact the employability of computer science graduates and develop a model for predicting employability. We start by using
exploratory factor analysis (EFA) to identify factors that affect the employability of computer science graduates, such as thinking and
emotional intelligence. Then we use confirmatory factor analysis (CFA) to validate and evaluate these factors obtained from EFA.
Additionally we create a two level model for employability prediction by combining based machine learning (ML) with generative
artificial intelligence (AI). In the level of prediction we utilize ML techniques, like random forest, k nearest neighbor, decision tree
and logistic regression. The performing model, from the stage is then used in the second stage of prediction. Here a generative multi in-one artificial neural network (GMA-NN) calculates the employability prediction. Finally, the study formulates a contingency matrix
for employability using the identified design factors and evaluates the model's performance and effectiveness using various design
metrics. Our results indicate that the LR+GMA-NN model we propose achieves the highest accuracy at 97.846%, surpassing the
existing state-of-the-art model by an impressive efficiency gain of 4.398%
